WO2022003407A1

WO2022003407A1 - Edge computing (ec) routing policies recommendation based on causal inference analytics

Info

Publication number: WO2022003407A1
Application number: PCT/IB2020/057083
Authority: WO
Inventors: Miguel Angel PUENTE PESTAÑA
Original assignee: Telefonaktiebolaget Lm Ericsson (Publ)
Priority date: 2020-06-30
Filing date: 2020-07-27
Publication date: 2022-01-06

Abstract

A network analytics node (100) implements a procedure for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In one aspect, the network analytics node receives (112, 192) a routing recommendation request from a policy control node (60). The routing recommendation request includes an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE. The network analytics node also determines (126, 194) a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request. The routing recommendation includes a Data Network Access Identifier (DNAI) that identifies one of an ECDN and a Central DN to which user traffic for the UE will be routed. Once determined, the network analytics node sends (128, 196) the routing recommendation to the policy control node.

Description

EDGE COMPUTING (EC) ROUTING POLICIES RECOMMENDATION BASED ON CAUSAL

INFERENCE ANALYTICS

RELATED APPLICATIONS

The present application claims priority to European Application No. 20382585.6, which was filed June 30, 2020, the disclosure of which is incorporated herein by reference in its entirety.

TECHNICAL FIELD

This application relates generally to routing user traffic in a communications network, and more particularly to determining routing recommendations for user traffic in an Edge Cloud Data Network (ECDN) or a Central DN based on a causal inference analysis.

BACKGROUND

Edge Computing

Edge computing is a distributed computing model that situates the computational and data storage resources of a communications network closer to where they are needed by the User Equipment (UEs) that use them. Typically, edge computing applies to non-roaming and Local Break Out (LBO) roaming scenarios. Regardless, however, situating the resources as such enables network operators and third-party service providers to be hosted close to a UE's point of attachment to the network (e.g., an Access Point - AP). This, in turn, allows for a more efficient delivery of services to a user, reduces the end-to-end latency, and reduces the load on the transport network.

There are a variety of standards documents related to support for edge computing. One such document is “5G; System Architecture for the 5G System” (3GPP TS 23.501 version 15.2.0 Release 15) dated June 2018. According to this document, a 5G Core Network selects a User Plane Function (UPF) close to the UE and executes the traffic steering from the UPF to the local Data Network (DN) via a N6 interface. The selection of a UPF can be based, for example, on the UE's subscription data, the UE’s current location, the information from an Application Function (AF), a policy, or other related traffic rules.

Additionally, because of user or AF mobility, service or session continuity may be required based on the requirements of the service or the 5G network. The 5G Core Network may also expose network information and capabilities to an Edge Computing Application Function. Further, depending on operator deployment, certain selected AFs can interact directly with the Control Plane Network Functions with which they need to interact. Other AFs, however, will need to use the external exposure framework via the Network Exposure Function (NEF).

According to TS 23.501 , edge computing can be supported by one or a combination of the following enabling functionalities. • User plane selection/reselection: The 5G Core Network can select/reselect a UPF to route the user traffic to a local DN;

• Local Routing and Traffic Steering: The 5G Core Network selects the traffic to be routed to the applications in the local DN. This includes the use of a single Packet Data Unit (PDU) Session with multiple PDU Session Anchor(s) (e.g., UL CL/IPv6 multi-homing).

• Session and service continuity to enable UE and AF mobility;

• An AF may influence UPF selection/reselection and traffic routing via a Policy Control Function (PCF) or NEF;

• Network capability exposure: 5G Core Network and AF to provide information to each other via a NEF or directly;

• Quality of Service (QoS) and Charging: The PCF provides rules for QoS Control and Charging for the traffic routed to the local DN;

• Support of Local Area Data Networks (LADN): The 5G Core Network provides support to connect to the LADN in a certain area where the applications are deployed.

Routing to Data Networks in 5GC

A DN Access Identifier (DNAI) identifies a user plane access to one or more DN(s) where applications are deployed. In edge computing, different ECDNs are identified by different DNAIs. Conventionally, in a 5GC data network, a Session Management Function (SMF) decides whether to apply traffic routing in a PDU Session. To accomplish this function, the SMF may use UL Classifier functionality or IPv6 multi-homing based on one or more DNAI(s) included in the Policy and Charging Control (PCC) rules. Additionally, an AF may send requests to influence the SMF routing decisions. More particularly, the AF requests may influence UPF selection/reselection and allow the SMF to route user traffic to a local access of a DN identified by a DNAI. An AF may also indicate the location of an application by means of a DNAI.

Causal inference

Humans often rationalize the world around them in terms of cause and effect. Understanding this relationship allows us to change our behavior and possibly improve future outcomes. Causal inference is a statistical tool that enables Artificial Intelligence (Al) and machine learning algorithms to reason in a similar manner. For example, changing the settings or parameters in a network may be the cause of higher (or lower) latency. Understanding how the changed settings or parameters affect network latency, then, allows network operators to select those settings and parameters that are most appropriate for the network.

Randomized Controlled Trials ( RCTs )

One standard for inferring cause and effect is randomized controlled trials (RCTs) or A/B tests. A RCT is a type of scientific (often medical) experiment that aims to reduce certain sources of bias when testing the effectiveness of new treatments. This is accomplished by randomly allocating subjects into two or more groups, treating each group of subjects differently, and then comparing the results of their respective treatments to those of a measured response.

More particularly, RCTs call for splitting a population of subjects (e.g., individuals) into at least two groups - a treatment/experimental group, to which a treatment is administered, and a control group to which nothing (or a placebo) is administered. The respective outcomes of the treatment and non-treatment for the groups is then measured. Provided the two groups are not too dissimilar, the effectiveness of the treatment can be inferred based on the difference in outcomes between the two groups.

EC and NWDAF

The 3GPP is currently studying in potential solutions for EC based on analytics in “3GPP TR 23.700-91 V0.3.0 (2020-01 )” dated January 2020. Although no solutions have yet been proposed, section 5.1 .6 of this document details a use case that is directed to the support of edge computing.

In more detail, the use case in TR 23.700-91 states that edge computing can improve the quality and experience of services by allowing huge data volume with low latency and high efficiency. Therefore, to support edge computing, 5GS introduced several enabling features, including giving the AF the ability to influence traffic routing, Local Area DN (LADN) functionality, and Uplink Classifier (UL CL)/Branching point functionality.

During the REL-16 eNA study, a use case known as "Use Case 7: NWDAF assisting 5G Edge Computing" was studied. Based on use case 7, a key issue was derived in supporting the User Plane (UP) path selection capabilities of the SMF and the PCF. This key issue is known as “Key Issue 6: NWDAF Assisting Traffic Routing.” However, the analytic information identified in use case 7 allowed the SMF to configure only internal UP paths.

According to use case 7, a Network Data Analytics Function (NWDAF) can enhance both edge computing and 5GS operations. Specifically, the NWDAF can be configured to assist with traffic influencing decisions and operations by providing analytics related to network and service data in advance. The use of the NWDAF in supporting edge computing is beneficial because it considers at least the following aspects:

• User and service mobility;

• Service/application characteristics, such as QoS/QoE;

• Optimal UP path determination; and

• Various edge computing deployment scenarios.

To support edge computing operations efficiently, 5GS could be evolved to support mobility management considering both UE and resource mobility/availability, traffic steering (e.g., UPF selection considering DNAI characteristics), UP path changes and optimization (e.g., SMF path allocation decisions and SSC mode selection), and the like. Consequently, it can improve the quality of hosted edge computing services. SUMMARY

According to the present disclosure, a network analytics node implements a procedure for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In one aspect, a method implemented at the network analytics node comprises receiving a routing recommendation request from a policy control node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE. The method further comprises determining a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed. Once determined, the method further comprises sending the routing recommendation to the policy control node.

In a second aspect of the present disclosure, a policy control node implements a method for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In this aspect, the method comprises sending a routing recommendation request to a network analytics node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE. The method then calls for receiving, in response, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed. Once received, the method calls for PCF 60 updating a routing policy for the UE based on the routing recommendation.

In a third aspect, the present disclosure provides a network analytics node for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In this aspect, the network analytics node comprises interface circuitry configured for communication with one or more network nodes in a communication network and processing circuitry operatively connected to the interface circuit. The processing circuitry is configured to receive a routing recommendation request from a policy control node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, determine a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and send the routing recommendation to the policy control node.

In a fourth aspect, the present disclosure provides a network analytics node for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In this aspect, the network analytics node is configured to receive a routing recommendation request from a policy control node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, determine a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and send the routing recommendation to the policy control node.

In a fifth aspect, the present disclosure provides a non-transitory computer-readable storage medium. The medium has a computer program comprising executable instructions stored thereon that, when executed by a processing circuit of a network analytics node in a communications network, causes the network analytics node to receive a routing recommendation request from a policy control node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, determine a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and send the routing recommendation to the policy control node.

In a sixth aspect, the present disclosure provides a policy control node for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In this aspect, the policy control node comprises interface circuitry configured for communication with one or more nodes in a communications network, and processing circuitry operatively connected to the interface circuitry. In this embodiment, the processing circuitry is configured to send a routing recommendation request to a network analytics node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, receive, in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and update a routing policy for the UE based on the routing recommendation.

In a seventh aspect, the present disclosure provides a policy control node for routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN. In this case, the policy control node is configured to send a routing recommendation request to a network analytics node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, receive, in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and update a routing policy for the UE based on the routing recommendation.

In an eighth aspect, the present disclosure provides a non-transitory computer-readable storage medium having a computer program comprising executable instructions stored thereon. When executed by processing circuitry in a policy control node in a communication network, the executable instructions of the computer program causes the policy control node to send a routing recommendation request to a network analytics node, wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE, receive, in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed, and update a routing policy for the UE based on the routing recommendation.

BRIEF DESCRIPTION OF THE DRAWINGS

Figure 1 illustrates an exemplary wireless communication network configured according to one embodiment of the present disclosure.

Figures 2A-2B are signaling diagrams illustrating a subscription-based method for determining routing recommendations according to one embodiment of the present disclosure.

Figure 3 is a signaling diagram illustrating a request-response based method for determining routing recommendations according to one embodiment of the present disclosure.

Figure 4 is a flow diagram illustrating a method, implemented at a network analytics node, for determining routing recommendations according to one embodiment of the present disclosure.

Figure 5 is a flow diagram illustrating a method, implemented at a policy control node, for determining routing recommendations according to one embodiment of the present disclosure.

Figure 6 illustrates an exemplary Network Data Analytics Function (NWDAF) node configured according to one embodiment of the present disclosure.

Figure 7 illustrates an exemplary Policy Control Function (PCF) node configured according to one embodiment of the present disclosure.

Figure 8 illustrates the main functional components of an exemplary Network Data Analytics Function (NWDAF) node configured according to one embodiment of the present disclosure.

Figure 9 illustrates the main functional components of an exemplary Policy Control Function (PCF) node configured according to one embodiment of the present disclosure. DETAILED DESCRIPTION

Referring now to the drawings, an exemplary embodiment of the disclosure will be described in the context of a 5G wireless communication network. Those skilled in the art will appreciate that the methods and apparatus herein described are not limited to use in 5G networks but may also be used in wireless communication networks operating according to other standards.

Figure 1 illustrates a wireless communication network 10 according to one exemplary embodiment. The wireless communication network 10 comprises a radio access network (RAN)

20 and a core network 30 employing a service-based architecture. The RAN 20 comprises one or more base stations 25 providing radio access to UEs 40 operating within the wireless communication network 10. The base stations 25 are also referred to as gNodeBs (gNBs). The core network 30 provides a connection between the RAN 20 and other packet data networks, such as the IMS or the Internet.

In one exemplary embodiment, the core network 30 comprises a plurality of network functions (NFs), such as a User Plane Function (UPF) 35, an Access And Mobility Management Function (AMF) 50, a Session Management Function (SMF) 55, a Policy Control Function (PCF) 60, a Unified Data Management (UDM) function 65, a Authentication Server Function (AUSF) 70, a Unified Data Repository (UDR) 75, a Network Exposure Function (NEF) 80, a Network Repository Function (NRF) 85, and a Network Slice Selection Function (NSSF) 90. The core network 30 additionally includes a NWDAF 100 for generating and distributing analytics reports. These NFs comprise logical entities that reside in one or more core network nodes, which may be implemented by one or more processors, hardware, firmware, or a combination thereof. The functions may reside in a single core network node or may be distributed among two or more core network nodes. The network 10 may further include one or more Application Functions (AFs) 95 providing services to the network and or subscribers. The AFs 95 may be located in the core network 30 or be external to the core network 30.

In conventional wireless communication network, the various NFs (e.g., SMF 55, AMF 50, etc.) in the core network 30 communicate with one another over predefined interfaces. In the service-based architecture shown in Figure 1 , instead of predefined interfaces between the control plane functions, the wireless communication network 10 uses a services model in which the NFs query the NRF 85 or other NF discovery node to discover and communicate with each other.

As stated above, the 3GPP is currently studying potential solutions for edge computing based on analytics in TR 23.700-91. However, each of these solutions face multiple challenges.

For example, with the currently proposed solutions, UEs are allocated/routed to the Edge Cloud (EC) in a non-intelligent manner. That is, the UEs are allocated to the EC based either on a static configuration, or in response to routing requests from an AF. Additionally, there is no guarantee that the currently proposed solutions fulfill the desired benefits of the EC. In more detail, the EC was initially conceived to fulfill certain objectives, such as to ensure low latency to latency critical applications, provide high throughput content from local servers, reduce the load on the core network (CN), and the like. The currently proposed solutions, however, simply assume that these objectives are fulfilled when the UE is allocated to an Edge Cloud. There is no close-loop feedback to ensure that the objectives are being fulfilled, or to trigger corrective actions if they are not being fulfilled.

Further, there may be situations in which there is no need to allocate a UE to the EC. Particularly, the QoS/QoE requirements for certain UEs can be fulfilled by Central DNs. In such cases, there is no need to allocate these UEs to the EC. However, the currently proposed solutions do not provide a mechanism to intelligently allocate and/or reallocate UEs among ECs and Central DNs when the QoS/QoE requirements of the UE can be fulfilled by the Central DN. That is, the currently proposed solutions do not allocate and/or reallocate UEs considering certain optimization objectives, such as fulfilling user QoE, reducing latency and/or the load on the CN, and the like.

Embodiments of the present disclosure, however, provide a system and method that enables intelligent EC routing decisions in 5GC networks. The embodiments are based on the analytics capabilities of the NWDAF and configure the NWDAF to:

• Provide EC routing recommendations (i.e. , DNAI recommendations) subject to predefined optimization objectives for the UEs and the applications that service the UEs. The optimization objectives can be, but are not limited to, fulfilling a user’s QoE/QoS requirements, minimizing communications latency, minimizing the load on the CN, balancing the traffic load among Edge Clouds and Central DNs, and the like. In these embodiments, Network Functions (NFs), such as the PCF, for example, can be configured to subscribe to the EC routing recommendations provided by NWDAF.

• Collect certain information to facilitate the intelligent EC routing recommendations. Particularly, the NWDAF is configured to obtain:

• The location of the UE (e.g., from the AMF);

• The location of the application(s) providing service to the UE (e.g., from the OSS);

• Network status and KPIs, such as information defining the state of network congestion, link capacities, and the like (e.g., from the UPF).

• Continuously monitor whether the optimization objective(s) is/are satisfied. In situations where one or more optimization objective(s) is/are not satisfied, the NWDAF is configured to execute causal inference analytics processes to determine a best Edge Cloud or Central DN that should serve a certain user and/or an application providing service to the user. In such embodiments:

• The NWDAF may optionally identify available DNs at a certain location based on information received from the OSS. If one of the available DNs are deemed suitable, and if an application of interest is not already deployed in that DN, the NWDAF may request the OSS to deploy that application in that DN.

• The NWDAF may initiate an “experiment procedure” to determine how a given EC routing policy would affect one or more UEs (i.e., how a given EC would perform for a given UE in a certain location and with a certain application). To perform the experiment procedures, the NWDAF first selects a set of one or more “test” UEs. The NWDAF then sends a request message to the PCF to route the set of “test” UEs to the DNAI identifying the given Edge Cloud.

• The NWDAF then measures the effect of the EC routing policy on the optimization objective(s) of the test UEs. If the NWDAF determines that the given EC routing policy is more appropriate for UEs in a given location executing a given application subject to the optimization objectives, the NWDAF will:

• Send a request message to the PCF requesting that the PCF to stop/cancel the experiment procedure; and

• Send the EC routing policy that was tested as an EC routing recommendation to the subscribed NFs. In this embodiment, the EC routing recommendation identifies the UE(s), the application, and the DNAI to which the recommendation applies.

It should be noted, however, that while the present embodiments may be implemented based on a subscription-based model, the present disclosure is not so limited. In other embodiments, which are described in more detail below, the NWDAF is configured to provide such EC routing recommendations responsive to receiving an explicit request for the EC routing recommendations.

The embodiments of the present disclosure therefore provide benefits and advantages that conventional systems cannot and/or do not provide. Among these benefits and advantages, the present embodiments:

• Intelligently allocate and/or reallocate UEs among ECs and Central DNs based on specific optimization objectives;

• Determine a “best” EC and/or Central DN for the UEs and the applications that service them based on a causal inference analytics procedure performed by the NWDAF; and

• Provide a dynamic and automated solution configured to respond adaptively to changing network conditions, such as the addition or removal of ECs, etc.

Subscription Model

Figures 2A-2B are flow diagrams illustrating a method 110 of determining routing recommendations for user traffic in an EC DN or a Central DN based on a causal inference analysis according to one embodiment of the present disclosure. In particular, Figures 2A-2B illustrate this embodiment as a subscription-based model in which PCF 60 subscribes to receive the EC routing recommendations from the NWDAF 100.

As seen in Figure 2A, method 110 begins with the PCF 60 subscribing to the NWDAF 100 for EC routing recommendations (line 112). In this embodiment, the subscription request includes a UE-ID (i.e., a unique identifier identifying the UE), an Application ID (i.e., a unique identifier identifying a given application to provide service to the UE), and an optimization objective for the UE that indicates a particular optimization that should be satisfied for the UE. Such optimizations may include, but are not limited to, fulfilling a user’s QoE/QoS requirements, minimizing communications latency, minimizing a load on the CN, balancing the traffic load among ECs and Central DNs, and the like.

Responsive to receiving the subscription request, the NWDAF 100 sends a request message to the AMF 50 requesting the UE’s location following standard data collection mechanisms (line 114). The AMF 50 then responds to the NWDAF 100 with the UE’s current location (line 116).

The NWDAF 100 then sends another request message to the OSS 105 requesting the locations of the applications identified by the Application-ID (line 118). In response, the OSS 105 sends a list of DNAIs identifying the DNs where the application is currently deployed (line 120).

The NWDAF 100 then sends a request message to the UPF 35 (or some other network entity) requesting the current status of the network(s) identified in the list of DNAIs using standard data collection mechanisms (line 122). The UPF 35 (or other requested entity) then responds with the network status information and the KPIs for the identified network(s) (line 124).

Once the NWDAF 100 has this information, the NWDAF 100 determines, based on the collected data and on the optimization objective(s), which of the DNAIs on the list of DNAIs is the best for the EC routing recommendation (box 126). The NWDAF 100 then sends an acknowledgment response to PCF 60 acknowledging the subscription request (line 128). The response includes the initial DNAI recommendation.

The NWDAF 100 then begins diagnostics for the optimization objective(s) based on the collected data and using standard data collection mechanisms (box 130). Upon the NWDAF 100 detecting that an optimization objective is not satisfied (box 132), the NWDAF 100 initiates an “experiment procedure” in which the NWDAF 100 determines an “experimental” or “test” EC routing recommendation to send to the PCF 60. This experimental EC routing recommendation will be evaluated to determine whether it will satisfy the optimization objective(s).

To accomplish this, the NWDAF 100 is configured to obtain a list of DNAIs. In one embodiment, for example, the NWDAF 100 may send a request message to the OSS 105 requesting that it provide the available DNs in the UE location (line 134). In some aspects, the request message may also request that OSS 105 provide a maximum distance that is allowed to exist between the UE and the DN. In response, the OSS 105 sends a response message to the NWDAF 100 including a list of available DNs (line 136). The response message also includes DNAI and location for each DN on the list.

In another embodiment, the NWDAF 100 sends a request message to the OSS 105 to deploy a given application in a specified DNAI, provided that the given application is not already deployed there (line 138). In these aspects, the OSS 105 is configured to respond with an acknowledgement message indicating whether the deployment request was accepted or not accepted (line 140).

The NWDAF 100 then decides to trigger a causal inference analytics procedure (line 142). To accomplish this function, the NWDAF 100 sends a message to the PCF 60 including an “experimental” EC routing policy to apply to temporarily route a group of UEs to an Edge Cloud. In these aspects, the NWDAF 100 may provide a list of UE-IDs identifying which particular UEs that have been selected for the experimental routing procedure, an Application-ID identifying the particular application the experiment pertains to, and a DNAI identifying the particular EC the UEs are to be routed towards for the experiment.

It should be noted that the present disclosure is not limited to triggering the causal inference analytics procedure using only a single Application-ID. In cases where the experimental EC routing recommendation targets multiple applications at the same time, the NWDAF 100 may send a list of Application-IDs identifying each application. Additionally, according to the present embodiments, this procedure may also be accomplished using the DNAI of a Central DN. Thus, the present embodiments are not limited only to providing routing recommendations for ECs. The PCF 60 then responds to the NWDAF 100 with an acknowledgement message including the list of UEs that are accepted for the experimental routing procedure (line 144), and triggers an EC routing policies update (box 146).

As seen in Figure 2B, the NWDAF 100 then begins measuring the effects of the EC routing policy on the optimization objective(s) of the UEs that were routed in accordance with the experimental EC routing recommendation (box 148). The NWDAF 100 then determines, based on the measured effects on the optimization objectives, that the experimental EC routing policy is more appropriate for UEs that are in a given location and executing a given application (box 150). This “experimental” EC routing recommendation is therefore the new routing recommendation. Responsive to this determination, the NWDAF 100 sends a message to the PCF 60 to cancel the experimental routing (line 152). The message may include the list of UE-IDs and Application-ID(s). Upon receipt of the message, PCF 60 returns the “temporarily routed” UEs back to their original routing policies (box 154), and acknowledges cancellation of the experimental routing to the NWDAF 100 (line 156). The NWDAF 100 then notifies the PCF 60, as well as any other subscribed network functions, of the new EC routing recommendation (line 158). In this aspect, the notification includes the UE-ID(s) and Application-ID(s) that the new recommendation pertains to as well as the recommended DNAI. The PCF 60 then accepts or rejects the recommendation (box 160) and responds to the NWDAF 100 indicating whether it accepted or rejected the recommendation (box 162).

Reauest/Resoonse Model

As previously stated, the present disclosure is not limited to utilizing a subscription-based model for providing EC routing recommendations. Rather, the NWDAF 100 is configured to provide such EC routing recommendations responsive to receiving an explicit request for the EC routing recommendations. This request/response model may be implemented in addition to or in lieu of the previously described subscription-based model.

Figure 3 is a flow diagram illustrating a method 170 of determining routing recommendations for user traffic in an EC DN or a Central DN based on a causal inference analysis, and using a request/response type of model according to one embodiment of the present disclosure. In particular, the NWDAF 100 activates the EC routing policies recommendation analytics and collects the corresponding data for the corresponding UEs, applications, and DNs (box 172). This can be accomplished internally by the NWDAF 100 using any method needed or desired; however, in one embodiment, this process is triggered by an OAM command.

At PDU session establishment, SMF 55 sends a request message to the PCF 60 requesting the PCC rules for a given UE-ID (line 174). The PCF 60 then sends a message to the NWDAF 100 requesting the EC routing recommendations (line 176). In this embodiment, the request comprises the UE-ID, the Application-ID(s), and the optimization objective for the UE. The NWDAF 100 then sends a response message to the PCF 60 including the recommended DNAI (line 178), which in turn, sends a response message to the SMF 55 comprising the Application-ID and the PCC rules and including the DNAI (line 180).

Figure 4 is a flow diagram illustrating an exemplary method 190 for determining routing recommendations for user traffic to one of an EC DN and a Central DN. In this embodiment, method 190 is implemented at a network analytics node, such as NWDAF 100 for example. However, this is for illustrative purposes only. Those of ordinary skill in the art should appreciate that method 190 may be implemented by any network node configured to perform network analytics, and in some cases, may be implemented by a network analytics system comprising multiple network analytics nodes.

As seen in Figure 4, network analytics node receives a routing recommendation request from a policy control node (box 192). In this embodiment, the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) and an application to provide service to the UE. Responsive to receiving the routing recommendation request, the network analytics node determines a routing recommendation based on the optimization objective (box 194). In accordance with the present embodiments, the routing recommendation may comprise a DNAI identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed. So determined, the network analytics node sends the routing recommendation to the policy control node (box 196).

The routing recommendation request in some embodiments further comprises an application ID that identifies the application to provide service to the UE.

In such embodiments, method 190 further comprises determining a list of DNAIs based on the application identifier.

In some embodiments, to determine a list of DNAIs based on the application identifier, the network analytics node sends an application location request to an operation support node, and receives, in return, the list of DNAIs from the operation support node. In these embodiments, the application location request sent to the operation support node includes the application ID received in the routing recommendation request.

In some embodiments, the routing recommendation request further comprises a UE ID that identifies the UE.

Some embodiments of method 190 further comprise determining a location of the UE based on the UE identifier, and determining a network status for the one of the EC DN and Central DN identified by the DNAI.

In some embodiments, the routing recommendation is also determined based on the UE location, the network status, and the list of DNAIs.

In some embodiments, the routing recommendation sent to the policy control node comprises a DNAI selected from the list of DNAIs.

In some embodiments, the routing recommendation is an initial routing recommendation. In these cases, the method 190 further comprises the network analytics node determining one or more updated routing recommendations based on diagnostic information associated with the optimization objective.

In some embodiments, method 190 further comprises the network analytics node determining that the optimization objective is not satisfied based on the diagnostic information.

In some embodiments, method 190 further comprises the network analytics node sending a message to the operations support node requesting a current location for the UE, a list of available DNs, and a maximum distance from the UE to each of the available DNs. In response, the network analytics node receives the list of available DNs from the operations support node. In this embodiment, each DN in the list includes a DNAI and a location.

Some embodiments of method 190 further comprise the network analytics node sending a request to the operations support node to deploy the application to a selected DNAI.

Some embodiments of method 190 further comprise the network analytics node performing a causal inference analysis at the network analytics node. In these embodiments, the causal inference analysis measures an effect of an updated routing policy on the optimization objective of each of one or more selected UEs. Some embodiments of method 190 further comprise the network analytics node performing the causal inference analysis by sending a message to the policy control node that causes the policy control node to initiate a routing policy update procedure. In these embodiments, the message comprises a list of candidate UE IDs on which to test the updated routing policy, a candidate application ID that identifies an application to be tested according to the updated routing policy, and a candidate DNAI identifying the network to which the updated routing policy should apply. Additionally, in some embodiments, the network analytics node receives an acknowledgement message from the policy control node in response. In these embodiments, the acknowledgement message comprises the list of one or more selected UEs.

Some embodiments of method 190 further comprise the network analytics node measuring the effect of the updated routing policy on the optimization objective of each of one or more selected UEs.

Some embodiments of method 190 further comprise the network analytics node determining a new routing recommendation for a selected UE based on the measured effects of the routing policy on the selected UE, indicating to the policy control node that the network analytics node has ended the causal inference analysis, and sending the new routing recommendation to the policy control node.

In some embodiments, the routing recommendation request from the policy control node is a subscription request.

In other embodiments, however, the routing recommendation request from the policy control node is an explicit request for the routing recommendation.

In some embodiments, the routing recommendation is an EC routing recommendation.

Figure 5 is a flow diagram illustrating an exemplary method 200 for determining routing recommendations for user traffic to one of an EC DN and a Central DN. In this embodiment, method 200 is implemented at a policy control node, such as PCF 60, for example.

As seen in Figure 5, method 200 comprises the PCF 60 sending a routing recommendation request to a network analytics node (box 202). The routing recommendation request comprises an optimization objective for one or both of a UE and an application to provide service to the UE. Method 200 also comprises the PCF 60 receiving, in response to the routing recommendation request, a routing recommendation for the UE (box 204). The routing recommendation comprises DNAI identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed. Method 200 then comprises PCF 60 updating a routing policy for the UE based on the routing recommendation (box 206).

Some embodiments of method 200 further comprise the policy control node 60 initiating a causal inference analytics procedure at the network analytics node in response to receiving a message from the network analytics node. The causal inference analysis measures an effect of a provisional routing policy on the optimization objective of each of one or more selected UEs. In some embodiments, the message from the network analytics node causes the policy control node to initiate a routing update policy procedure, and comprises the policy control node 60 a list of candidate UE IDs on which to test the provisional routing policy, a candidate application ID that identifies an application to be tested according to the provisional routing policy, and a candidate DNAI identifying one of a target ECDN and a target Central DN to which the provisional routing policy should apply.

Some embodiments of method 200 further comprise the policy control node 60 initiating the causal inference analytics procedure at the network analytics node by sending an acknowledgement message to the network analytics node, wherein the acknowledgement message comprises a list of one or more selected UEs on which the network analytics node should perform the causal inference analytics procedure, and temporarily updating a current routing policy of the one or more selected UEs to the provisional routing policy.

Some embodiments of method 200 further comprise the policy control node 60 restoring the current routing policy of the one or more selected UEs responsive to receiving an indication from the network analytics node that the causal inference analysis procedure is complete.

In some embodiments, the routing recommendation is an initial routing recommendation. In these aspects method 200 further comprises the policy control node 60 receiving an updated routing recommendation for the one or more selected UEs, and updating the routing policy of the given UE according to the updated routing recommendation. In these cases, the updated routing recommendation comprises the candidate DNAI, the candidate application ID, and the UE ID of a given UE to which the updated routing recommendation applies.

An apparatus can perform any of the methods herein described by implementing any functional means, modules, units, or circuitry. In one embodiment, for example, the apparatuses comprise respective circuits or circuitry configured to perform the steps shown in the method figures. The circuits or circuitry in this regard may comprise circuits dedicated to performing certain functional processing and/or one or more microprocessors in conjunction with memory. For instance, the circuitry may include one or more microprocessor or microcontrollers, as well as other digital hardware, which may include Digital Signal Processors (DSPs), special-purpose digital logic, and the like. The processing circuitry may be configured to execute program code stored in memory, which may include one or several types of memory such as read-only memory (ROM), random-access memory, cache memory, flash memory devices, optical storage devices, etc. Program code stored in memory may include program instructions for executing one or more telecommunications and/or data communications protocols as well as instructions for carrying out one or more of the techniques described herein, in several embodiments. In embodiments that employ memory, the memory stores program code that, when executed by the one or more processors, carries out the techniques described herein. Figure 6 illustrates a network node configured to function as a network analytics node, such as NWDAF 100, according to one embodiment of the present disclosure. As seen in Figure 6, the NWDAF 100 comprises interface circuitry 210, processing circuitry 212, and memory circuitry 214.

The communication circuitry 210 comprises network interface circuitry for communicating with other network nodes in the wireless communication network over a communication network. Such nodes include, but are not limited to, a policy control node (e.g., PCF 60), an operations support node (e.g., OSS 105), an access and mobility management function (e.g., AMF 50), and a user plane function (e.g., UPF 35).

Processing circuitry 212 controls the overall operation of the NWDAF 100 and is configured to perform one or more of the methods 110 and 190 shown in Figures 2A-2B and 4, respectively. Such processing includes coding and modulation of transmitted data signals, and the demodulation and decoding of received data signals. The processing circuitry 212 may comprise one or more microprocessors, hardware, firmware, or a combination thereof.

Memory circuitry 214 comprises both volatile and non-volatile memory for storing computer program code and data needed by the processing circuitry 212 for operation. Memory circuitry 214 may comprise any tangible, non-transitory computer-readable storage medium for storing data including electronic, magnetic, optical, electromagnetic, or semiconductor data storage. Memory circuitry 214 stores a computer program 216 comprising executable instructions that configure the processing circuitry 212 to implement the methods 110 and 190 shown in Figures 2A-2B and 4, respectively. A computer program in this regard may comprise one or more code modules corresponding to the means or units described above. In general, computer program instructions and configuration information are stored in a non-volatile memory, such as a ROM, erasable programmable read only memory (EPROM) or flash memory. Temporary data generated during operation may be stored in a volatile memory, such as a random access memory (RAM). In some embodiments, computer program 216 for configuring the processing circuitry 212 as herein described may be stored in a removable memory, such as a portable compact disc, portable digital video disc, or other removable media. The computer program 216 may also be embodied in a carrier such as an electronic signal, optical signal, radio signal, or computer readable storage medium.

Figure 7 illustrates a network node configured to function as a policy control node, such as PCF 60, according to one embodiment of the present disclosure. As seen in Figure 7, the PCF 60 comprises interface circuitry 220, processing circuitry 222, and memory circuitry 224. As described in more detail later, a computer program 226 that configures PCF 60 to operate according to the present embodiments may be stored in memory circuitry 224.

The communication circuitry 220 comprises network interface circuitry for communicating with other network nodes in the wireless communication network over a communication network. Such nodes include, but are not limited to, a network analytics node (e.g., NWDAF 100), an operations support node (e.g., OSS 105), an access and mobility management function (e.g., AMF 50), and a user plane function (e.g., UPF 35).

Processing circuitry 222 controls the overall operation of PCF 60 and is configured to perform one or more of the methods 170 and 200 shown in Figures 3 and 5, respectively. Such processing may include, for example, the coding and modulation of transmitted data signals, and the demodulation and decoding of received data signals. The processing circuitry 222 may comprise one or more microprocessors, hardware, firmware, or a combination thereof.

Memory circuitry 224 comprises both volatile and non-volatile memory for storing computer program code and data needed by the processing circuitry 222 for operation. Memory circuitry 224 may comprise any tangible, non-transitory computer-readable storage medium for storing data including electronic, magnetic, optical, electromagnetic, or semiconductor data storage. Memory circuitry 224 stores computer program 226 comprising executable instructions that configure the processing circuitry 222 to implement the methods 170 and 200 shown in Figures 3 and 5, respectively. A computer program in this regard may comprise one or more code modules corresponding to the means or units described above. In general, computer program instructions and configuration information are stored in a non-volatile memory, such as a ROM, erasable programmable read only memory (EPROM) or flash memory. Temporary data generated during operation may be stored in a volatile memory, such as a random access memory (RAM). In some embodiments, computer program 226 for configuring the processing circuitry 222 as herein described may be stored in a removable memory, such as a portable compact disc, portable digital video disc, or other removable media. The computer program 226 may also be embodied in a carrier such as an electronic signal, optical signal, radio signal, or computer readable storage medium.

Figure 8 illustrates an exemplary network analytics node, such as NWDAF 100, configured to perform methods 110 and 190 shown in Figures 2A-2B and 4, respectively. The NWDAF 100 comprises a communications unit/module 230, a routing recommendation unit/module 232, and a causal inference analytics unit/module 234. The various units/modules 230-234 can be implemented by hardware and/or by software code that is executed by a processor or processing circuitry 212.

The communications unit/module 230 is configured to receive routing recommendation requests from other network entities, such as PCF 60. The requests may be explicit requests for EC routing recommendations, or may be subscription requests to subscribe to receiving EC routing recommendations from NWDAF 100. Additionally, the communications unit/module 230 is configured to send the EC routing recommendations to the PCF 60.

The routing recommendation unit/module 232 performs its functions responsive to receiving the routing recommendation request from PCF 60, and is configured to determine the EC routing recommendations based on an optimization objective received with the request. The EC routing recommendations comprise a DNAI that identifies one of an EC DN and a Central DN to which user traffic for the UE will be routed.

The causal inference analytics unit/module 234 is configured implement an analytics process in which NWDAF 100 initiates the “experiment procedure” describe previously to determine how a given EC routing policy would affect one or more UEs. In more detail, the causal inference analytics unit/module 234 selects a set of one or more “test” UEs and triggers PCF 60 into routing traffic from the set of “test” UEs to a selected DNAI. The causal inference analytics unit/module 234 then measures the effect of the EC routing policy on the optimization objective(s) of the test UEs, and based on those measurements, sends the “test” routing procedure to the PCF 60 as a new EC routing recommendation the NFs that are subscribed to receive such information.

Figure 9 illustrates an exemplary policy control node, such as PCF 60, configured to perform methods 170 and 200 shown in Figures 3 and 5, respectively. The PCF 60 comprises a communications unit/module 240 and a routing policy unit/module 242. As above, the various units/modules 240 and 242 can be implemented by hardware and/or by software code that is executed by a processor or processing circuitry 222.

The communications unit/module 240 is configured to receive routing recommendations from NWDAF 100, as well as other data and information from NWDAF 100 and other network entities. In addition, the communications unit/module 240 is configured to send requests to NWDAF 100 to provide EC routing recommendations. As previously described, the may be explicit requests for EC routing recommendations, or may be subscription requests to subscribe to receiving EC routing recommendations from NWDAF 100. The routing policy unit/module 242 is configured to update its policy tables according to the EC routing recommendations received from NWDAF 100.

Those skilled in the art will also appreciate that embodiments herein further include corresponding computer programs. A computer program comprises instructions which, when executed on at least one processor of an apparatus, cause the apparatus to carry out any of the respective processing described above. A computer program in this regard may comprise one or more code modules corresponding to the means or units described above.

Embodiments further include a carrier containing such a computer program. This carrier may comprise one of an electronic signal, optical signal, radio signal, or computer readable storage medium.

In this regard, embodiments herein also include a computer program product stored on a non-transitory computer readable (storage or recording) medium and comprising instructions that, when executed by a processor of an apparatus, cause the apparatus to perform as described above.

Embodiments further include a computer program product comprising program code portions for performing the steps of any of the embodiments herein when the computer program product is executed by a computing device. This computer program product may be stored on a computer readable recording medium.

The present invention may, of course, be carried out in other ways than those specifically set forth herein without departing from essential characteristics of the invention. The present embodiments are to be considered in all respects as illustrative and not restrictive, and all changes coming within the meaning and equivalency range of the appended claims are intended to be embraced therein.

Claims

CLAIMS What is claimed is:

1 . A method (110, 190), implemented at a network analytics node (100), for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the method comprising: receiving (112, 192) a routing recommendation request from a policy control node (60), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; determining (126, 194) a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and sending (128, 196) the routing recommendation to the policy control node.

2. The method of claim 1 wherein the routing recommendation request further comprises an application ID that identifies the application to provide service to the UE.

3. The method of claim 2 further comprising determining a list of DNAIs based on the application identifier.

4. The method of claims 2-3 wherein determining a list of DNAIs based on the application identifier comprises: sending (118) an application location request to an operation support node (105), wherein the request includes the application ID received in the routing recommendation request; and receiving (120) the list of DNAIs from the operation support node.

5. The method of any of claims 1 -4 wherein the routing recommendation request further comprises a UE ID that identifies the UE.

6. The method of any of claims 1 -5 further comprising: determining (114, 116) a location of the UE based on the UE identifier; determining (122, 124) a network status for the one of the ECDN and Central DN identified by the DNAI.

7. The method of any of claims 1 -6 wherein the routing recommendation is also determined based on the UE location, the network status, and the list of DNAIs.

8. The method of claim 7 wherein the routing recommendation sent to the policy control node comprises a DNAI selected from list of DNAIs.

9. The method of any of claims 1 -8 wherein the routing recommendation is an initial routing recommendation, and further comprising determining one or more updated routing recommendations based on diagnostic information associated with the optimization objective.

10. The method of claim 8 further comprising determining (132) that the optimization objective is not satisfied based on the diagnostic information.

11 . The method of claim 8 further comprising: sending (134) a message to the operations support node requesting a current location for the UE, a list of available DNs, and a maximum distance from the UE to each of the available DNs; and receiving (136) the list of available DNs from the operations support node, wherein each DN in the list includes a DNAI and a location.

12. The method of claim 8 further comprising sending (138) a request to the operations support node to deploy the application to a selected DNAI.

13. The method of any of claims 1 -12 further comprising performing a causal inference analysis at the network analytics node, wherein the causal inference analysis measures an effect of an updated routing policy on the optimization objective of each of one or more selected UEs.

14. The method of claim 13 wherein performing the causal inference analysis comprises: sending (142) a message to the policy control node that causes the policy control node to initiate a routing policy update procedure, wherein the message comprises: a list of candidate UE IDs on which to test the updated routing policy; a candidate application ID that identifies an application to be tested according to the updated routing policy; and a candidate DNAI identifying the network to which the updated routing policy should apply; and receiving (144), in response, an acknowledgement message from the policy control node comprising the list of one or more selected UEs.

15. The method of claims 13-14 further comprising measuring (148) the effect of the updated routing policy on the optimization objective of each of one or more selected UEs.

16. The method of claims 13-15 further comprising: determining (150) a new routing recommendation for a selected UE based on the measured effects of the routing policy on the selected UE; indicating (152) to the policy control node that the network analytics node has ended the causal inference analysis; and sending (158) the new routing recommendation to the policy control node.

17. The method of any of the preceding claims wherein the routing recommendation request from the policy control node is a subscription request.

18. The method of claim 1 wherein the routing recommendation request from the policy control node is an explicit request for the routing recommendation.

19. The method of any of the preceding claims wherein the routing recommendation is an EC routing recommendation.

20. A method (170, 200), implemented at a policy control node (60), for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the method comprising: sending (176, 202) a routing recommendation request to a network analytics node (100), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; receiving (178, 204), in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and updating (206) a routing policy for the UE based on the routing recommendation.

21 . The method of claim 20 further comprising initiating a causal inference analytics procedure at the network analytics node responsive to receiving (142) a message from the network analytics node, wherein the causal inference analysis measures an effect of a provisional routing policy on the optimization objective of each of one or more selected UEs.

22. The method of claim 21 wherein the message from the network analytics node causes the policy control node to initiate a routing update policy procedure and comprises: a list of candidate UE IDs on which to test the provisional routing policy; a candidate application ID that identifies an application to be tested according to the provisional routing policy; and a candidate DNAI identifying one of a target ECDN and a target Central DN to which the provisional routing policy should apply.

23. The method of claims 21-22 wherein initiating the causal inference analytics procedure at the network analytics node comprises: sending (144) an acknowledgement message to the network analytics node, wherein the acknowledgement message comprises a list of one or more selected UEs on which the network analytics node should perform the causal inference analytics procedure; and temporarily updating (146) a current routing policy of the one or more selected UEs to the provisional routing policy.

24. The method of claims 20-23 further comprising restoring (154) the current routing policy of the one or more selected UEs responsive to receiving (152) an indication from the network analytics node that the causal inference analysis procedure is complete.

25. The method of claims 20-24 wherein the routing recommendation is an initial routing recommendation, and further comprising: receiving (158) an updated routing recommendation for the one or more selected UEs, wherein the updated routing recommendation comprises: the candidate DNAI; the candidate application ID; and the UE ID of a given UE to which the updated routing recommendation applies; and updating (160) the routing policy of the given UE according to the updated routing recommendation.

26. A network analytics node (100) for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the network analytics node comprising: interface circuitry (210) configured for communication with one or more network nodes in a communication network; and processing circuitry (212) operatively connected to the interface circuit and configured to: receive (112, 192) a routing recommendation request from a policy control node (60), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; determine (126, 194) a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and send (128, 196) the routing recommendation to the policy control node.

27. The network analytics node of claim 26 wherein the routing recommendation request further comprises an application ID that identifies the application to provide service to the UE.

28. The network analytics node of claim 27 wherein the processing circuitry is further configured to determine a list of DNAIs based on the application identifier.

29. The network analytics node of claims 27-28 wherein to determine the list of DNAIs based on the application identifier, the processing circuitry is configured to: send (118) an application locations request to an operation support node (105), wherein the request includes the application ID received in the routing recommendation request; and receive (120) the list of DNAIs from the operation support node.

30. The network analytics node of any of claims 26-29 wherein the routing recommendation request further comprises a UE ID that identifies the UE.

31 . The network analytics node of any of claims 26-30 wherein the processing circuitry is configured to: determine (114, 116) a location of the UE based on the UE identifier; determine (122, 124) a network status for the one of the ECDN and Central DN identified by the DNAI.

32. The network analytics node of any of claims 26-31 wherein the processing circuitry is further configured to determine the routing recommendation based on the UE location, the network status, and the list of DNAIs.

33. The network analytics node of claim 32 wherein the routing recommendation sent to the policy control node comprises a DNAI selected from list of DNAIs.

34. The network analytics node of any of claims 26-33 wherein the routing recommendation is an initial routing recommendation, and wherein the processing circuitry is further configured to determine one or more subsequent routing recommendations based on feedback associated with the optimization objective.

35. The network analytics node of claim 33 wherein the processing circuitry is further configured to determine (132) that the optimization objective is not satisfied based on diagnostic information obtained by the network analytics node.

36. The network analytics node of claim 33 wherein the processing circuitry is further configured to: send (134) a message to the operations support node requesting a current location for the UE, a list of available DNs, and a maximum distance from the UE to each of the available DNs; and receive (136) the list of available DNs from the operations support node, wherein each DN in the list includes a DNAI and a location.

37. The network analytics node of claim 33 wherein the processing circuitry is further configured to send (138) a request to the operations support node to deploy the application to a selected DNAI.

38. The network analytics node of any of claims 26-37 wherein the processing circuitry is further configured to perform a causal inference analysis measuring an effect of an updated routing policy on the optimization objective of each of one or more selected UEs.

39. The network analytics node of claim 38 wherein to perform the causal inference analysis, the processing circuitry is configured to: send (142) a message to the policy control node that causes the policy control node to initiate a routing policy update procedure, wherein the message comprises: a list of candidate UE IDs on which to test the updated routing policy; a candidate application ID that identifies an application to be tested according to the updated routing policy; and a candidate DNAI identifying the network to which the updated routing policy should apply; and receive (144), in response, an acknowledgement message from the policy control node comprising the list of one or more selected UEs.

40. The network analytics node of claims 38-39 wherein the processing circuitry is further configured to measure (148) the effect of the updated routing policy on the optimization objective of each of one or more selected UEs.

41 . The network analytics node of claims 38-40 wherein the processing circuitry is further configured to: determine (150) a new routing recommendation for a selected UE based on the measured effects of the routing policy on the selected UE; indicate (152) to the policy control node that the network analytics node has ended the causal inference analysis; and send (158) the new routing recommendation to the policy control node.

42. The network analytics node of any of claims 26-41 wherein the routing recommendation request from the policy control node is a subscription request.

43. The network analytics node of claim 26 wherein the routing recommendation request from the policy control node is an explicit request for the routing recommendation.

44. The network analytics node of any of claims 26-43 wherein the routing recommendation is an EC routing recommendation.

45. A network analytics node (100) for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the network analytics node being configured to: receive (112, 192) a routing recommendation request from a policy control node (60), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; determine (126, 194) a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and send (128, 196) the routing recommendation to the policy control node.

46. A non-transitory computer-readable storage medium (214) having a computer program (216) comprising executable instructions stored thereon that, when executed by a processing circuit (212) of a network analytics node (100) in a communications network (10), causes network analytics node to: receive (112, 192) a routing recommendation request from a policy control node (60), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; determine (126, 194) a routing recommendation based on the optimization objective responsive to receiving the routing recommendation request, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and send (128, 196) the routing recommendation to the policy control node.

47. A policy control node (60) for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the policy control node comprising: interface circuitry (220) configured for communication with one or more nodes in a communications network; and processing circuitry (222) operatively connected to the interface circuitry and configured to: send (176, 202) a routing recommendation request to a network analytics node (100), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; receive (178, 204), in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and update (206) a routing policy for the UE based on the routing recommendation.

48. The policy control node of claim 46 wherein the processing circuitry is further configured to initiate a causal inference analytics procedure at the network analytics node, wherein the causal inference analysis measures an effect of a provisional routing policy on the optimization objective of each of one or more selected UEs.

49. The policy control node of claim 48 wherein to initiate the causal inference analytics procedure at the network analytics node, the processing circuitry is configured to: receive (142) a message from the network analytics node, wherein the message causes the policy control node to initiate a routing update policy procedure and comprises: a list of candidate UE IDs on which to test the provisional routing policy; a candidate application ID that identifies an application to be tested according to the provisional routing policy; and a candidate DNAI identifying one of a target ECDN and a target Central DN to which the provisional routing policy should apply; determine a list of one or more selected UEs on which the network analytics node should perform the causal inference analytics procedure; send (144) an acknowledgement message to the network analytics node, wherein the acknowledgement message comprises the list of one or more selected UEs; and temporarily update (146) a current routing policy of the one or more selected UEs to the provisional routing policy.

50. The policy control node of claims 47-49 wherein the processing circuitry is further configured to restore (154) the current routing policy of the one or more selected UEs responsive to receiving (152) an indication from the network analytics node that the causal inference analysis procedure is complete.

51 . The policy control node of claims 47-50 wherein the routing recommendation is an initial routing recommendation, and wherein the processing circuitry is further configured to: receive (158) an updated routing recommendation for the one or more selected UEs, wherein the updated routing recommendation comprises: the candidate DNAI; the candidate application ID; and the UE ID of a given UE to which the updated routing recommendation applies; and update (160) the routing policy of the given UE according to the updated routing recommendation.

52. A policy control node (60) for determining routing recommendations for user traffic to one of an Edge Cloud Data Network (ECDN) and a Central DN, the policy control node being configured to: send (176, 202) a routing recommendation request to a network analytics node (100), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; receive (178, 204), in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and update (206) a routing policy for the UE based on the routing recommendation.

53. A non-transitory computer-readable storage medium (224) having a computer program (226) comprising executable instructions stored thereon that, when executed by processing circuitry (222) in a policy control node (60) in a communication network (10), causes the policy control node to: send (176, 202) a routing recommendation request to a network analytics node (100), wherein the routing recommendation request comprises an optimization objective for one or both of a User Equipment (UE) (40) and an application to provide service to the UE; receive (178, 204), in response to the routing recommendation request, a routing recommendation for the UE, wherein the routing recommendation comprises a Data Network Access Identifier (DNAI) identifying one of an ECDN and a Central DN to which user traffic for the UE will be routed; and update (206) a routing policy for the UE based on the routing recommendation.