hash stringlengths 32 32 | doc_id stringlengths 7 13 | section stringlengths 3 121 | content stringlengths 0 2.2M |
|---|---|---|---|
e341f05abb94e155b675f29c6779693c | 104 230 | 10.5.5 Required data | For querying the A1 policy identifiers, the rApp optionally provides an A1 policy type identifier and/or a Near RT RIC identifier as a query parameter. The A1 policy functions respond with the list of A1 policy identifiers. 10.6 A1-related use case 4: Create, Query, Update and Delete an A1 policy |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.1 Overview | This use case enables rApps to create an A1 policy, query an A1 policy, update an A1 policy, and delete an A1 policy. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.2 Background and goal of the use case | Create A1 policy, Query A1 policy, Update A1 policy, and Delete A1 policy procedures are defined as part of A1 policy management service in R1GAP [1]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.3 Entities/resources involved in the use case | 1) A1 policy functions: a) support functionality to allow rApps to request creating7, querying, updating and deleting an A1 policy. 2) rApp: a) initiates the procedures for creating an A1 policy, querying an A1 policy, updating an A1 policy, and deleting an A1 policy. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 94 |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.4.1 Create a single A1 policy in an identified Near-RT RIC | Table 10.6.4.1-1: Create a single A1 policy in an identified Near-RT RIC Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp initiates creation of an A1 policy in an identified Near-RT RIC. Actors and Roles - rApp in the role of the A1 policy management service Consumer. - A1 policy functions in the role of the A1 policy management service Producer. Assumptions rApp is authorized to consume the A1 policy management service. Preconditions The rApp is aware about the A1 policy type. Begins when The rApp determines to create a new A1 policy. Step 1 (M) The rApp requests the A1 policy functions to create a new A1 policy of certain type by providing a policy type identifier, policy information, rApp identifier, and a Near-RT RIC identifier. Step 2 (M) The A1 policy functions validate if the rApp is authorized to create A1 policies. Step 3 (M) The A1 policy functions generate the A1 policy identifier. Step 4 (M) The A1 policy functions create a single A1 policy in the Near-RT RIC corresponding to the given Near-RT RIC identifier. See A1UCR [6], clause 6.3 Step 5 (M) The A1 policy functions store the mapping of A1 policy identifier to the Near-RT RIC identifier. Step 6 (M) The A1 policy functions respond to rApp with the A1 policy identifier of the created A1 policy. Ends when The rApp has created a new A1 policy. Exceptions n/a Post Conditions An A1 policy exists and the rApp will be able to query, update and delete the A1 policy. Traceability REQ-R1-A1P-FUN9. @startuml 'https://plantuml.com/sequence-diagram' !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "A1 policy functions " as A1P endbox rApp -> A1P :1 <<R1>> Create A1 policy request\n (policy type identifier, policy information, Near-RT RIC identifier, rAppId) activate A1P A1P --> A1P :2 AuthZ note right Check authorization in collaboration with SME functions end note A1P --> A1P : 3 Generate A1 policy identifier ref over A1P 4 create single policy(see A1 UCR clause 6.3) end ref A1P --> A1P :5 Store A1 policy identifier A1P -> rApp : 6 <<R1>> Create A1 policy response (policy identifier) Deactivate A1P @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 95 Figure 10.6.4.1-1: Create a single A1 policy in an identified Near-RT RIC flow diagram 10.6.4.2 Query an A1 policy Table 10.6.4.2-1: Query an A1 policy Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp queries an existing A1 policy. Actors and Roles - rApp in the role of an A1 policy management service Consumer. - A1 policy functions in the role of an A1 policy management service Producer. Assumptions The rApp is authorized to consume the A1 policy management service. Preconditions - The rApp is aware about the A1 policy identifier of the A1 policy that it wants to query. - A1 policy functions are aware of the Near-RT RIC identifier corresponding to this A1 policy identifier so that it can query the A1 policy at the relevant Near-RT RIC instance. Begins when The rApp determines to query an existing A1 policy. Step 1 (M) The rApp requests the A1 policy functions to query an existing A1 policy by providing the A1 policy identifier, and its rApp identifier. Step 2 (M) The A1 policy functions validate if the rApp is authorized to query an A1 policy. Step 3 (M) A1 policy functions respond with the result of the A1 policy query including policy information. See note Ends when The rApp has the A1 policy information. Exceptions n/a Post Conditions n/a Traceability REQ-R1-A1P-FUN9. NOTE: The A1 policy functions may use A1UCR [6], clause 6.4.3.2 Query single policy to retrieve the A1 policy information from the relevant Near-RT RIC whenever an rApp queries an A1 policy. It may also choose to respond to the rApp query with the latest policy information previously received from the relevant Near-RT RIC. @startuml 'https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid ETSI ETSI TS 104 230 V10.0.0 (2026-02) 96 autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box "Non-RT RIC FWK/SMO FWK" #cadetBlue participant "A1 policy functions " as A1P rApp -> A1P :<<R1>>Query A1 policy request (A1 policy identifier, \n rAppId) activate A1P A1P --> A1P : AuthZ note right Check authorization in collaboration with SME functions end note A1P -> rApp :<<R1>>Query A1 policy response (policy information) Deactivate A1P @enduml Figure 10.6.4.2-1: Query A1 policy use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 97 |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.4.3 Update an A1 policy | Table 10.6.4.3-1: Update an A1 policy Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp updates an existing A1 policy. Actors and Roles - rApp in the role of an A1 policy management service Consumer. - A1 policy functions in the role of an A1 policy management service Producer. Assumptions The rApp is authorized to consume the A1 policy management service. Preconditions - The rApp is aware of the identifier of the A1 policy that it wants to update. - A1 policy functions are aware of the existing policy to be updated was created by the rApp. - A1 policy functions are aware of the Near-RT RIC identifier corresponding to this A1 policy identifier so that it can update the A1 policy at the relevant Near-RT RIC instance. Begins when The rApp determines to update an existing A1 policy. Step 1 (M) The rApp requests the A1 policy functions to update an existing A1 policy by providing the A1 policy identifier, policy update information, rApp identifier. Step 2 (M) The A1 policy functions validate if the rApp is authorized to update an A1 policy. Step 3 (M) The A1 policy functions update a single A1 policy in the Near-RT RIC. See A1UCR [6], clause 6.5 Step 4 (M) The A1 policy functions respond with the result of the A1 policy update. Ends when The rApp has updated the A1 policy. Exceptions n/a Post Conditions The content of the A1 policy has changed. Traceability REQ-R1-A1P-FUN9. @startuml 'https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box "Non-RT RIC FWK/SMO FWK" #cadetBlue participant "A1 policy functions " as A1P rApp -> A1P :<<R1>>Update A1 policy request (A1 policy identifier, \n policy information, rApp Id) activate A1P A1P --> A1P : AuthZ note right Check authorization in collaboration with SME functions end note ref over A1P 3 Update single policy(see A1 UCR clause 6.5) end ref autonumber 4 A1P -> rApp :<<R1>>Update A1 policy response deactivate A1P @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 98 Figure 10.6.4.3-1: Update A1 policy use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.4.4 Delete an A1 policy | Table 10.6.4.4-1: Delete an A1 policy Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp deletes an existing A1 policy. Actors and Roles - rApp in the role of an A1 policy management service Consumer. - A1 policy functions in the role of an A1 policy management service Producer. Assumptions The rApp is authorized to consume A1 policy management service. Preconditions - The rApp is aware of the identifier of the A1 policy that it wants to delete. - A1 policy functions are aware of the existing policy to be deleted was created by the rApp. - A1 policy functions are aware of the Near-RT RIC identifier corresponding to this A1 policy identifier so that it can delete the A1 policy at the relevant Near-RT RIC instance. Begins when The rApp determines to delete an existing A1 policy. Step 1 (M) The rApp requests the A1 policy functions to delete an existing A1 policy by providing the A1 policy identifier and rApp identifier. Step 2 (M) The A1 policy functions validate if the rApp is authorized to delete an A1 policy. Step 3 (M) The A1 policy functions delete a single A1 policy in the Near-RT RIC. See A1UCR [6], clause 6.6 Step 4 (M) The A1 policy functions respond with the result of A1 policy delete. Ends when The rApp has deleted an A1 policy. Exceptions n/a Post Conditions The A1 policy has ceased to exist. Traceability REQ-R1-A1P-FUN9. @startuml 'https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box "Non-RT RIC FWK/SMO FWK" #cadetBlue participant "A1 policy functions " as A1P endbox ETSI ETSI TS 104 230 V10.0.0 (2026-02) 99 rApp -> A1P :<<R1>>Delete A1 policy request (A1 policy identifier, \n rAppId) activate A1P A1P --> A1P : AuthZ note right Check authorization in collaboration with SME functions end note ref over A1P 3 Delete single policy(see A1 UCR clause 6.6) end ref autonumber 4 A1P -> rApp :<<R1>>Delete A1 policy response Deactivate A1P @enduml Figure 10.6.4.4-1: Delete A1 policy use case flow diagram. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.6.5 Required data | For creating an A1 policy of a certain type in a Near-RT RIC, the rApp provides an A1 policy type identifier, A1 policy information, its rApp identifier (rAppId) and, a Near-RT RIC identifier. The A1 policy functions respond with the A1 policy identifier of the created policy. For querying an A1 policy, the rApp provides an A1 policy identifier, and its rApp identifier (rAppId). The A1 policy functions respond with A1 policy information. For updating an A1 policy, the rApp provides an A1 policy identifier, A1 policy updated information, and its rApp identifier (rAppId). For deleting an A1 policy, the rApp provides an A1 policy identifier and its rApp identifier (rAppId). |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7 A1-Related use case 5: Query the status of an A1 policy | |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.1 Overview | This use case provides the description and requirements for querying the status of an A1 policy status. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.2 Background and goal of the use case | Query A1 policy status procedure is defined as part of A1 policy management service in R1GAP [1]. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 100 |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.3 Entities/resources involved in the use case | 1) A1 policy functions: a) support functionality to allow rApps to query the status of an A1 policy. 2) rApp: a) initiates the procedure for querying status of an A1 policy. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.4.1 Query the status of an A1 policy | Table 10.7.4.1-1: Query status of A1 policy Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp is informed about the status of an A1 policy. Actors and Roles - rApp in the role of A1 policy management service Consumer. - A1 policy functions in the role of A1 policy management service Producer. Assumptions rApp is authorized to use A1 policy management service. Preconditions rApp is aware of the identifier of the A1 policy that it wants to get the status for. Begins when The rApp determines the need to query the status of an A1 policy. Step 1 (M) The rApp queries the status of an A1 policy with A1 policy functions by providing an rAppId and A1 policy identifier. Step 2 (M) The A1 policy functions validate if the rApp is authorized to query the status of an A1 policy. Step 3 (M) The A1 policy functions respond with the status of the corresponding A1 policy. See note Ends when The rApp has received the status of an A1 policy. Exceptions n/a Post Conditions The rApp has the status of an A1 policy. Traceability REQ-R1-A1P-FUN5. NOTE: The A1 policy functions may use A1UCR [6], clause 6.7.3.1 Query policy status, or A1UCR [6], clause 6.7.3.2 Notify policy status change to respond with the latest status of the corresponding A1 policy. @startuml 'https://plantuml.com/sequence-diagram' !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box "Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "A1 policy functions" as A1P endbox rApp -> A1P: <<R1>>Query A1 policy status request(rAppId,A1 policy identifier) activate A1P A1P --> A1P :AuthZ note right Check authorization in collaboration with SME functions end note A1P -> rApp : <<R1>>Query A1 policy status response Deactivate A1P @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 101 Figure 10.7.4.1-1: Query status of A1 policy use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.7.5 Required data | For the Query A1 policy status request, the rApp provides the rAppId and the A1 policy identifier. The A1 policy functions respond with the status of the A1 policy. The status information of A1 policy is defined in A1 TD [3]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8 A1-Related use case 6: Subscribe to status of an A1 policy | |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.1 Overview | This use case provides the description and requirements for subscribe to status of an A1 policy. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.2 Background and goal of the use case | Subscribe to A1 policy status changes procedure is defined as part of A1 policy management service in R1GAP [1]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.3 Entities/resources involved in the use case | 1) A1 Policy functions in the role of A1 policy management service Producer: a) support functionality to allow rApps to subscribe to and unsubscribe from A1 policy status changes; b) support notifying the rApp regarding the changes in the status of an A1 policy; c) provide response of success or failure to subscribe to A1 policy status changes and unsubscribe to A1 policy status changes. 2) rApp: a) supports functionality to initiate requests to subscribe to and unsubscribe from A1 policy status changes; b) management service Producer. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 102 |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.4.1 Subscribe to A1 policy status change | Table 10.8.4.1-1: subscribe to A1 policy status change Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp subscribes to notifications about changes of the status of an A1 policy. Actors and Roles - rApp in the role of A1 policy management service Consumer. - A1 policy functions in the role of A1 policy management service Producer. Assumptions rApp is aware of the A1 policy identifier of the A1 policy that it is interested in subscribing to status change notifications for. Preconditions rApp is authorized to consume the A1 policy management service. Begins when The rApp determines the need to subscribe to notifications, regarding changes in the status of an A1 policy. Step 1 (M) The rApp request the A1 policy functions to subscribe to notifications regarding changes in the status of an A1 policy by providing the rAppId and A1 policy identifier. Step 2 (M) The A1 policy functions check whether the rApp is authorized to send a subscription request. Step 3 (M) The A1 policy functions respond with subscription identifier. Ends when The rApp was able to subscribe to notifications. Exceptions n/a Post Conditions The rApp can receive notifications when there are any changes in the status of the A1 policy. The rApp can unsubscribe from status change notifications of the A1 policy. Traceability REQ-R1-A1P-FUN6. @startuml 'https://plantuml.com/sequence-diagram' !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "A1 policy functions "as A1P endbox rApp -> A1P : <<R1>> subscribe A1 policy status change request\n(rAppId, A1 policy identifier) activate A1P A1P --> A1P : AuthZ note right Check authorization in collaboration with SME functions end note A1P -> rApp : <<R1>> subscribe A1 policy status change response (subscription identifier) Deactivate A1P @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 103 Figure 10.8.4.1-1: Subscribe to A1 policy status change use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.4.2 Notify A1 policy status change | Table 10.8.4.2-1: Notify A1 policy status change use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp receives notifications about changes in the status of an A1 policy. Actors and Roles - rApp in the role of A1 policy management service Consumer. - A1 policy functions in the role of A1 policy management service Producer. Assumptions n/a Preconditions The rApp has subscribed to notifications on changes in the status of an A1 policy. Begins when The A1 policy functions determine to send notifications, regarding changes in the status of the A1 policy to the subscribed rApp. Step 1 (M) The A1 policy functions detect changes in status (i.e. an A1 policy status becomes enforced or not enforced). Step 2 (M) The A1 policy functions send notifications regarding the changes to the subscribed rApp with subscription identifier and information about status changes of the A1 policy. Ends when The rApp was able to process the changes in the status of the A1 policy. Exceptions n/a Post Conditions n/a Traceability REQ-R1-A1P-FUN7. @startuml ‘https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box "Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "A1 policy functions " as A1P endbox autonumber A1P→A1P: detect the status change A1P-> rApp : <<R1>> notify changes \n(subscription identifier) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 104 Figure 10.8.4.2-1: Notify status change use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.4.3 Unsubscribe from A1 policy status change | Table 10.8.4.3-1: Unsubscribe from A1 policy status change Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp unsubscribes from A1 policy status change notifications. Actors and Roles - rApp in the role of A1 policy management service Consumer. - A1 policy functions in the role of A1 policy management service Producer. Assumptions n/a Preconditions - The rApp is authorized to unsubscribe from A1 policy status change notifications. - The rApp has subscribed to A1 policy status change notifications. Begins when The rApp determines to unsubscribe from A1 policy status change. Step 1 (M) The rApp requests the A1 policy functions to unsubscribe from A1 policy status changes with rAppId and subscription identifier. Step 2 (M) The A1 policy functions check whether the rApp is authorized to unsubscribe. Step 3 (M) The A1 policy functions send a response. Ends when The rApp was able to unsubscribe from A1 policy status change. Exceptions n/a Post Conditions The rApp is not subscribed to A1 policy status change. Traceability REQ-R1-A1P-FUN9. @startuml 'https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "A1 policy functions " as A1P endbox autonumber rApp -> A1P : <<R1>> Unsubscribe from A1 policy status change request(rAppId,subscription identifier) activate A1P A1P -> A1P : AuthZ note right Check authorization in collaboration with SME functions end note ETSI ETSI TS 104 230 V10.0.0 (2026-02) 105 A1P -> rApp : <<R1>> Unsubscribe from A1 policy status change response deactivate A1P @enduml Figure 10.8.4.3-1: Unsubscribe from A1 policy status change use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 10.8.5 Required data | To receive notifications regarding changes in status change of an A1 policy, the rApp as A1 policy service Consumer subscribes to notifications by providing rAppId and A1 policy identifier. When the A1 policy functions have successfully processed the request, they respond by providing a subscription identifier. The A1 policy functions send notifications to the subscribed rApp when there are changes in status changes of the A1 policy by providing subscription identifier, and information about the changes (i.e. A1 policy status becomes enforced or not enforced). For unsubscribing from notifications, a subscribed rApp needs to provide the rAppId and subscription identifier. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11 Use cases for AI/ML Workflow Services | 11.1 AI/ML workflow-related use case 1: AI/ML model registration and deregistration |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.1 Overview | This use case enables an rApp as AI/ML model management and exposure services Consumer to register AI/ML models, query AI/ML model registration, update AI/ML model registrations and deregister AI/ML models, |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.2 Background and goal of the use case | An AI/ML model producer rApp can register an AI/ML model, query and update the registration of an AI/ML model and deregister an AI/ML model with the AI/ML workflow functions. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.3 Entities/resources involved in the use case | 1) AI/ML workflow functions as AI/ML model management and exposure services Producer: a) support the functionality to allow an rApp to register, query and update the registration of, and deregister an AI/ML model; b) provide the response of success or failure results to the register AI/ML model, query, and update AI/ML model registration, and deregister AI/ML model request. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 106 2) rApp as AI/ML model management and exposure services Consumer: a) support to initiate the procedure to register an AI/ML model, query and update the registration of an AI/ML model and deregister an AI/ML model. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.4.1 Register AI/ML model | Table 11.1.4.1-1: Register AI/ML model Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML model producer rApp registers an AI/ML model. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions The AI/ML model that they intend to register in AI/ML workflow functions has not been registered before. Preconditions The rApp is authorized to access the AI/ML model management and exposure services for registering an AI/ML model. The AI/ML model is certified and onboarded in SMO run time library. Begins when The AI/ML model producer rApp determines the need to register an AI/ML model. Step 1 (M) The rApp requests the AI/ML workflow functions to register the AI/ML model by providing the rAppId and AI/ML model-related information. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to register an AI/ML model. Step 3 (M) The AI/ML workflow functions validate the register AI/ML model request. Step 4 (M) The AI/ML workflow functions register the AI/ML model. Step 5 (M) The AI/ML workflow functions respond to rApp with successful AI/ML model registration along with AI/ML model registration identifier. Ends when The AI/ML model producer rApp was able to register the AI/ML model. Exceptions n/a Post Conditions The AI/ML model is discoverable, and the rApp can query, update, or delete the model registration. Traceability REQ-R1-AIML-Registermodel-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as src endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox src -> ML: <<R1>> Register AI/ML model request (rAppId, AI/ML model-related information) ML -> ML: Authz note right Check authorization in collaboration with SME functions end note ML -> ML: Validate ETSI ETSI TS 104 230 V10.0.0 (2026-02) 107 ML -> ML: Register AI/ML model ML -> src: <<R1>> Register AI/ML model response (AI/ML model registration identifier) @enduml Figure 11.1.4.1-1: Register AI/ML model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.4.2 Query AI/ML model registration | Table 11.1.4.2-1: Query AI/ML model registration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML model producer rApp queries an AI/ML model registration. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services for registering an AI/ML model. - The AI/ML model has been registered by the rApp. Begins when The AI/ML model producer rApp determines the need to query the registration of a registered AI/ML model. Step 1 (M) The AI/ML model producer rApp requests to query the registration of an AI/ML model by providing rAppId and registration identifier. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to query the AI/ML model registration. Step 3 (M) The AI/ML workflow functions look up the information of queried AI/ML model. Step 4 (M) The AI/ML workflow functions respond to rApp with AI/ML model- related information. Ends when The AI/ML model producer rApp was able to query the AI/ML model registration. Exceptions n/a Post Conditions The AI/ML model is discoverable, and the rApp can query, update or delete the AI/ML model registration. Traceability REQ-R1-AIML-Registermodel-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent ETSI ETSI TS 104 230 V10.0.0 (2026-02) 108 autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as src endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox src -> ML: <<R1>> Query AI/ML model request (rAppId, AI/ML model registration identifier) ML -> ML: Authz note right Check authorization in collaboration with SME functions end note ML -> ML: Look up AI/ML model registration ML -> src: <<R1>> Query AI/ML model response (AI/ML model-related information) @enduml Figure 11.1.4.2-1: Query AI/ML model registration use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 109 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.4.3 Update AI/ML model registration | Table 11.1.4.3-1: Update AI/ML model registration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML model producer rApp updates the registration of an AI/ML model. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services for updating an AI/ML model registration. - The AI/ML model has been registered by the rApp. Begins when The AI/ML model producer rApp determines the need to update the registration of an AI/ML model. Step 1 (M) The AI/ML model producer rApp requests to update the registration of an AI/ML model by providing rAppId, updated model registration information, etc. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to update the registration. Step 3 (M) The AI/ML workflow functions validate the update AI/ML model registration request. Step 4 (M) The AI/ML workflow functions update the registration of the AI/ML model. Step 5 (M) The AI/ML workflow functions respond to the model producer rApp with successful update AI/ML model registration response. Ends when The AI/ML model producer rApp was able to update the registration of the AI/ML model. Exceptions n/a Post Conditions The rApp can query, update, or delete the AI/ML model registration. The updated AI/ML model is discoverable and is notified to all subscribed model consumer rApps. Traceability REQ-R1-AIML-Registermodel-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as app endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber app -> aif: <<R1>> Update model registration request \n (rAppId, AI/ML model registration identifier, \n updated model registration information) activate aif aif --> aif: AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Validate aif --> aif: Update model \n registration information app <- aif: <<R1>> Update model registration response deactivate aif @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 110 Figure 11.1.4.3-1: Update registration of an AI/ML model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.4.4 Deregister AI/ML model | Table 11.1.4.4-1: Deregister AI/ML model Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML model producer rApp deregisters an AI/ML model. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services. - The AI/ML model has been registered by the rApp. Begins when The AI/ML model producer rApp determines the need to deregister an AI/ML model it has previously registered. Step 1 (M) The rApp requests AI/ML workflow functions to deregister the AI/ML model by providing rAppId and AI/ML model registration identifier. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to deregister the AI/ML model. Step 3 (M) The AI/ML workflow functions delete the registration of the AI/ML model. Step 4 (M) The AI/ML workflow functions respond to the rApp with successful deregister AI/ML model response. Ends when The AI/ML model producer rApp was able to deregister the AI/ML model. Exceptions n/a Post Conditions The AI/ML model is not discoverable any longer. Traceability REQ-R1-AIML-Registermodel-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid ETSI ETSI TS 104 230 V10.0.0 (2026-02) 111 skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as src endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox src -> ML: <<R1>> Deregister AI/ML model request \n(rAppId, AI/ML model registration identifier) ML -> ML: Authz note right Check authorization in collaboration with SME functions end note ML -> ML: Deregister AI/ML model ML -> src: <<R1>> Deregister AI/ML model response @enduml Figure 11.1.4.4-1: Deregister AI/ML model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.1.5 Required data | The AI/ML model-related information includes the AI/ML model identifier, AI/ML model description, input, and output datatype, etc. For registering an AI/ML model, the rApp provides the rAppId and AI/ML model-related information. On successful registration, AI/ML workflow functions provide an AI/ML model registration identifier for the registration of the AI/ML model. For querying the registration of an AI/ML model, the rApp needs to provide the rAppId and the AI/ML model registration identifier. For updating the registration of an AI/ML model, the rApp needs to provide the rAppId, the AI/ML model registration identifier and the updated AI/ML model-related information or the modified part of the AI/ML model-related information. For deregistering an AI/ML model, the rApp provides the rAppId and the AI/ML model registration identifier. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 112 11.2 AI/ML workflow-related use case 2: Discovery of AI/ML Model |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.1 Overview | This use case enables an rApp to retrieve information about registered AI/ML Models and their associated information. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.2 Background and goal of the use case | The discover AI/ML model procedure is defined as part of the AI/ML workflow services in R1GAP [1]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.3 Entities/resources involved in the use case | 1) AI/ML workflow functions in the role of AI/ML model management and exposure service Producer: a) support functionality allowing rApps to discover the registered AI/ML Models; 2) rApp in the role of AI/ML model management and exposure Consumer: a) initiates the procedure to discover the registered AI/ML models. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.4.1 Discover AI/ML models | Table 11.2.4.1-1: Discover AI/ML models Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp discovers registered AI/ML models. Actors and Roles - rApp in the role of AI/ML model management and exposure Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML workflow services. - At least one AI/ML model is registered with the AI/ML workflow functions. Begins when The rApp determines the need to discover the registered AI/ML models. Step 1 (M) The rApp requests the AI/ML workflow functions for the information on the AI/ML Models by providing rAppId and optional selection criteria. Step 2 (M) The AI/ML workflow functions validate if the rApp is authorized to discover the registered AI/ML models. Step 3 (M) The AI/ML workflow functions provide the information about registered AI/ML models. The list contains the information of those AI/ML models that match the filtering criteria if those were provided by the rApp, or of all AI/ML models otherwise. For each AI/ML model, the model identifier and model metadata are provided. Ends when The rApp was able to discover the registered AI/ML models. Exceptions n/a Post Conditions The rApp can use the AI/ML model identifier in other AI/ML workflow services. Traceability REQ-R1-AI/ML-discovery-FUN1. @startuml 'https://plantuml.com/sequence-diagram !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 ETSI ETSI TS 104 230 V10.0.0 (2026-02) 113 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant " AI/ML workflow functions " as aif endbox rApp ->aif:<<R1>> Discover AI/ML model request\n(rAppId,Selection criteria) activate aif aif --> aif:AuthZ note right Check authorization in collaboration with SME functions end note aif -> rApp :<<R1>>Discover AI/ML model response\n(model identifiers,model metadata) deactivate aif @enduml Figure 11.2.4.1-1: Discover AI/ML model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.2.5 Required data | For the Discover AI/ML model request, the rApp provides the rAppId and optional selection criteria. The AI/ML workflow functions respond with information which includes the model identifiers and metadata information. The metadata information contains AI/ML model related information and training related information. If selection criteria information was provided in the related request, the response only contains information about those AI/ML models information that match the filtering criteria. 11.3 AI/ML workflow-related use case 3: AI/ML training - AI/ML workflow functions producing AI/ML training services |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.1 Overview | This use case defines how an rApp as an AI/ML training services Consumer requests AI/ML training, queries AI/ML training job status, cancels AI/ML training, and receives notification about AI/ML training job status change, when the AI/ML workflow functions in the SMO/Non-RT RIC framework acts as the AI/ML training services Producer. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 114 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.2 Background and goal of the use case | An AI/ML training services Consumer rApp can request the AI/ML workflow functions in the Non-RT RIC framework to train an AI/ML model. For on-going AI/ML training, the AI/ML training services Consumer rApp can query the training job status or cancel the AI/ML training. If the training job status is changed, the AI/ML workflow functions can notify the AI/ML training service Consumer rApp about the training job status change. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.3 Entities/resources involved in the use case | 1) AI/ML workflow functions as AI/ML training services Producer: a) support the functionality to allow an AI/ML training services Consumer rApp to request training, query training job status, and cancel training; b) provide the response of success or failure results to the request training, query training job status, and cancel training request; c) support the functionality to notify an AI/ML training services Consumer rApp about the training job status change. 2) AI/ML training services Consumer rApp: a) support to initiate the procedure to request AI/ML training, query AI/ML training job status, and cancel AI/ML training; b) support the functionality to receive notification of AI/ML training job status change. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.4.1 Request AI/ML training | Table 11.3.4.1-1: Request AI/ML training use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp requests training of an AI/ML model. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. Begins when The AI/ML training services Consumer rApp determines the need to train an AI/ML model. Step 1 (M) The AI/ML training services Consumer rApp requests the AI/ML workflow functions to train an AI/ML model providing rAppId, information for training, and optionally a notification URI, etc. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML training services Consumer rApp is authorized to request training. Step 3 (M) The AI/ML workflow functions validate the request. Step 4 (M) The AI/ML workflow functions create the training job. Step 5 (M) The AI/ML workflow functions respond to the AI/ML training services Consumer rApp with training job identifier as a parameter. Ends when The AI/ML training services Consumer rApp is able to create the training job at the AI/ML workflow functions. Exceptions n/a Post Conditions The AI/ML workflow functions can retrieve model to be trained from model repository and consume training data from DME. Traceability REQ-R1-AI/ML-training-FUN1. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 115 @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber con -> aif: <<R1>> Request training request \n (rApp id, information for training, \n notification URI, etc.) activate aif aif --> aif: AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Validate aif --> aif: Create training job con <- aif: <<R1>> Request training response \n (training job id) deactivate aif @enduml Figure 11.3.4.1-1: Request AI/ML training use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 116 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.4.2 Query AI/ML training job status | Table 11.3.4.2-1: Query AI/ML training job status use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp query training job status of a created training job. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. Begins when The AI/ML training services Consumer rApp determines the need to query the training job status. Step 1 (M) The AI/ML training services Consumer rApp queries the AI/ML workflow functions about the status of a created training job by providing rAppId and training job identifier. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML training services Consumer rApp is authorized to query the training job status. Step 3 (M) The AI/ML workflow functions validate the request. Step 4 (M) The AI/ML workflow functions respond to the AI/ML training services Consumer rApp with training job status. Ends when The AI/ML training services Consumer rApp is able to obtain the training job status. Exceptions n/a Post Conditions The training job status is known to the AI/ML training services Consumer rApp. Traceability REQ-R1-AI/ML-training-FUN2. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber con -> aif: <<R1>> Query training status request \n (rApp id, training id) activate aif aif --> aif: AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Validate con <- aif: <<R1>> Query training status response \n (training job status) deactivate aif @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 117 Figure 11.3.4.2-1: Query AI/ML training job status use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.4.3 Cancel AI/ML training | Table 11.3.4.3-1: Cancel AI/ML training use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp cancels training of an AI/ML model. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. Begins when The AI/ML training services Consumer rApp determines to cancel the training of an AI/ML model. Step 1 (M) The AI/ML training services Consumer rApp cancels the training by providing rAppId and training job identifier. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML training services Consumer rApp is authorized to cancel the training. Step 3 (M) The AI/ML workflow functions validate the request. Step 4 (M) The AI/ML workflow functions stop the training job. Step 5 (M) The AI/ML workflow functions respond to the AI/ML training services Consumer rApp. Ends when The AI/ML workflow functions are able to cancel training of an AI/ML model. Exceptions n/a Post Conditions The training is cancelled, and the training job is terminated. Traceability REQ-R1-AI/ML-training-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent ETSI ETSI TS 104 230 V10.0.0 (2026-02) 118 box Non-RT RIC #whitesmoke box #ivory participant AI/ML training services \n Consumer rApp as con endbox box Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber con -> aif: <<R1>> Cancel training request \n (rApp id, training job id.) activate aif aif → aif: AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Validate aif --> aif: Stop training job con <- aif: <<R1>> Cancel training response deactivate aif @enduml Figure 11.3.4.3-1: Cancel AI/ML training use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 119 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.4.4 Notify AI/ML training job status change | Table 11.3.4.4-1: Notify AI/ML training job status change use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML workflow functions notify the AI/ML training services Consumer rApp about the training job status change of a created training job. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. - The AI/ML training services Consumer rApp had provided a notification URI to the AI/ML workflow functions. Begins when The AI/ML workflow functions determine the need to notify training job status change to the AI/ML training services Consumer rApp. Step 1 (M) The AI/ML workflow functions notify the training job status change to AI/ML training services Consumer rApp. Ends when The AI/ML training services Consumer rApp is notified about the updated training job status. Exceptions n/a Post Conditions The updated training job status is known to the AI/ML training services Consumer rApp. Traceability REQ-R1-AI/ML-training-FUN3. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber con <- aif: <<R1>> Notify training job status change \n (training job id, training job status, model id) @enduml Figure 11.3.4.4-1: Notify AI/ML training job status change use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 120 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.3.5 Required data | For requesting training, the AI/ML training services Consumer rApp needs to provide its rAppId, information for training (e.g. information about training dataset, model identifier, training criteria.), and optionally a notification URI to receive notification about training job status change. If the AI/ML workflow functions accept the training request, they assign a training job identifier for the created training job. For querying training job status, the AI/ML training services Consumer rApp needs to provide its rAppId and the training job identifier. The AI/ML workflow functions respond with the status of the training job identified by the training job identifier. For cancelling training, the AI/ML training services Consumer rApp needs to provide its rAppId and the training job identifier. For notifying training job status change, the AI/ML workflow functions provide the training job status, the training job identifier, and optionally model identifier. 11.4 AI/ML workflow-related use case 4: AI/ML training - rApp producing AI/ML training services |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.1 Overview | This use case defines how an rApp as an AI/ML training services Consumer requests AI/ML training, queries AI/ML training job status, cancels AI/ML training, and receives notification about AI/ML training job status change, when another rApp acts as the AI/ML training services Producer. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.2 Background and goal of the use case | An AI/ML training services Consumer rApp can request the AI/ML training service Producer rApp to train an AI/ML model. For on-going AI/ML training, the AI/ML training services Consumer rApp can query the training job status or cancel the AI/ML training. If the training job status is changed, the AI/ML training services Producer rApp can notify the AI/ML training service Consumer rApp about the training job status change. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.3 Entities/resources involved in the use case | 1) AI/ML training services Producer rApp: a) supports the functionality to allow an AI/ML training services Consumer rApp to request training, query training job status, and cancel training; b) provides the response of success or failure results to the request training, query training job status, and cancel training request; c) supports the functionality to notify an AI/ML training services Consumer rApp about the training job status change. 2) AI/ML training services Consumer rApp: a) supports to initiate the procedure to request AI/ML training, query AI/ML training job status, and cancel AI/ML training; b) supports the functionality to receive notification of AI/ML training job status change. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 121 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.4.1 Request AI/ML training | Table 11.4.4.1-1: Request AI/ML training use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp requests training of an AI/ML model. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - AI/ML training service Producer rApp in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. - The AI/ML training services Producer rApp is deployed, authenticated, and authorized to produce AI/ML training services. Begins when The AI/ML training services Consumer rApp determines the need to train an AI/ML model. Step 1 (M) The AI/ML training services Consumer rApp requests the AI/ML training services Producer rApp to train an AI/ML model providing rAppId, information for training, and optionally a notification URI, etc. Step 2 (M) The AI/ML training services Producer rApp checks with SME functions whether the AI/ML training services Consumer rApp is authorized to request training. Step 3 (M) The AI/ML training services Producer rApp validates the request. Step 4 (M) The AI/ML training services Producer rApp creates the training job. Step 5 (M) The AI/ML training services Producer rApp responds to the AI/ML training services Consumer rApp with training job identifier as a parameter. Ends when The AI/ML training services Consumer rApp is able to create the training job at the AI/ML training services Producer rApp. Exceptions n/a Post Conditions The AI/ML training service Producer rApp can retrieve model to be trained from model repository and consume training data from DME. Traceability REQ-R1-AI/ML-training-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con participant "AI/ML training services \n Producer rApp" as aif endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "SME functions" as sme endbox endbox autonumber con -> aif: Request training request \n (rApp id, information for training, \n notification URI, etc.) activate aif aif <-> sme: <<R1>> Check authorization in \n collaboration with SME functions aif --> aif: Validate aif --> aif: Create training job con <- aif: Request training response \n (training job id) deactivate aif @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 122 Figure 11.4.4.1-1: Request AI/ML training use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.4.2 Query AI/ML training job status | Table 11.4.4.2-1: Query AI/ML training job status use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp query training job status of a created training job. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - AI/ML training service Producer rApp in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. - The AI/ML training services Producer rApp is deployed, authenticated, and authorized to produce AI/ML training services. Begins when The AI/ML training services Consumer rApp determines the need to query the training job status. Step 1 (M) The AI/ML training services Consumer rApp queries the AI/ML training service Producer rApp about the status of a created training job by providing rAppId and training job identifier. Step 2 (M) The AI/ML training service Producer rApp checks with SME functions whether the AI/ML training services Consumer rApp is authorized to query the training job status. Step 3 (M) The AI/ML training service Producer rApp validates the request. Step 4 (M) The AI/ML training service Producer rApp responds to the AI/ML training services Consumer rApp with training job status. Ends when The AI/ML training services Consumer rApp is able to obtain the training job status. Exceptions n/a Post Conditions The training job status is known to the AI/ML training services Consumer rApp. Traceability REQ-R1-AI/ML-training-FUN2. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 123 @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con participant "AI/ML training services \n Producer rApp" as aif endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "SME functions" as sme endbox endbox autonumber con -> aif: Query training job status request (rApp id, training id) activate aif aif <-> sme: <<R1>> Check authorization in \n collaboration with SME functions aif --> aif: Validate con <- aif: Query training job status response (training job status) deactivate aif @enduml Figure 11.4.4.2-1: Query AI/ML training job status use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 124 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.4.3 Cancel AI/ML training | Table 11.4.4.3-1: Cancel AI/ML training use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training services Consumer rApp cancels training of an AI/ML model. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - AI/ML training service Producer rApp in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. - The AI/ML training services Producer rApp is deployed, authenticated, and authorized to produce AI/ML training services. Begins when The AI/ML training services Consumer rApp determines to cancel the training of an AI/ML model. Step 1 (M) The AI/ML training services Consumer rApp cancels the training by providing rAppId and training job identifier. Step 2 (M) The AI/ML training service Producer rApp checks with SME functions whether the AI/ML training services Consumer rApp is authorized to cancel the training. Step 3 (M) The AI/ML training service Producer rApp validates the request. Step 4 (M) The AI/ML training service Producer rApp stops the training job. Step 5 (M) The AI/ML training service Producer rApp responds to the AI/ML training services Consumer rApp. Ends when The AI/ML training service Producer rApp is able to cancel training of an AI/ML model. Exceptions n/a Post Conditions The training is cancelled, and the training job is terminated. Traceability REQ-R1-AI/ML-training-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML training services \n Consumer rApp" as con participant "AI/ML training services \n Producer rApp" as aif endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "SME functions" as sme endbox endbox autonumber con -> aif: Cancel training request \n (rApp id, training job id.) activate aif aif <-> sme: <<R1>> Check authorization in \n collaboration with SME functions aif --> aif: Validate aif --> aif: Stop training job con <- aif: Cancel training response deactivate aif @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 125 Figure 11.4.4.3-1: Cancel AI/ML training use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.4.4 Notify AI/ML training job status change | Table 11.4.4.4-1: Notify AI/ML training job status change use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML training service Producer notifies the AI/ML training services Consumer rApp about the training job status change of a created training job. Actors and Roles - AI/ML training services Consumer rApp in the role of Service Consumer. - AI/ML training service Producer rApp in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML training services Consumer rApp is deployed, authenticated, and authorized to consume AI/ML training services. - The AI/ML training services Consumer rApp had provided a notification URI to the AI/ML training services Producer rApp. - The AI/ML training services Producer rApp is deployed, authenticated, and authorized to produce AI/ML training services. Begins when The AI/ML training service Producer determines the need to notify training job status change to the AI/ML training services Consumer rApp. Step 1 (M) The AI/ML training service Producer notifies the training job status change to AI/ML training services Consumer rApp. Ends when The AI/ML training services Consumer rApp is notified about the updated training job status. Exceptions n/a Post Conditions The updated training job status is known to the AI/ML training services Consumer rApp. Traceability REQ-R1-AI/ML-training-FUN3. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent ETSI ETSI TS 104 230 V10.0.0 (2026-02) 126 skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke participant "AI/ML training services \n Consumer rApp" as con participant "AI/ML training services \n Producer rApp" as aif endbox autonumber con <- aif: Notify training job status change \n (training job id, training job status, model id) @enduml Figure 11.4.4.4-1: Notify AI/ML training job status change use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.4.5 Required data | For requesting training, the AI/ML training services Consumer rApp needs to provide its rAppId, information for training (e.g. information about training dataset, model identifier, training criteria.), and optionally a notification URI to receive notification about training job status change. If the AI/ML training service Producer rApp accepts the training request, it assigns a training job identifier for the created training job. For querying training job status, the AI/ML training services Consumer rApp needs to provide its rAppId and the training job identifier. The AI/ML training service Producer rApp responses with the status of the training job identified by the training job identifier. For cancelling training, the AI/ML training services Consumer rApp needs to provide its rAppId and the training job identifier. For notifying training job status change, the AI/ML training service Producer rApp provides the training job status, the training job identifier, and optionally model identifier. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5 AI/ML workflow-related use case 5: AI/ML model Retrieve | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.1 Overview | This use case defines how an rApp retrieves model location details as a consumer of registered AI/ML model(s). |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.2 Background and goal of the use case | The AI/ML model retrieve procedures are defined as part of the AI/ML workflow services as part of AI/ML workflow services in R1GAP [1]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.3 Entities/resources involved in the use case | 1) AI/ML workflow functions: a) support model management and exposure functionality to allow an rApp to retrieve the location details of AI/ML model(s). ETSI ETSI TS 104 230 V10.0.0 (2026-02) 127 2) rApp: a) initiates the procedure to retrieve the location details of AI/ML model(s). |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.4.1 Retrieve model | Table 11.5.4.1-1: Model retrieve Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp retrieves AI/ML model(s). Actors and Roles - rApp in the role of Model retrieve service Consumer. - AI/ML workflow functions in the role of Model retrieve service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML Model retrieve service. - The rApp has the model identifier(s). Begins when The rApp determines the need to retrieve the AI/ML model. Step 1 (M) The rApp requests the AI/ML workflow functions to retrieve an AI/ML model by providing the rApp ID, and the AI/ML model identifier(s). Step 2 (M) The AI/ML workflow functions check with SME functions whether the rApp is authorized to retrieve AI/ML models. Step 3 (M) The AI/ML workflow functions fetch the model location details of registered AI/ML model(s). Step 4 (M) The AI/ML workflow functions respond to rApp AI/ML model identifier, model location details. Ends when The rApp has the location details. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-Retrievemodel-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "Model Consumer rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox group Model Consumer rApp model retrieve rApp -> ML: <<R1>> Retrieve AI/ML model information request \n (rAppId, AI/ML model identifier(s)) ML -> ML: AuthZ ML -> ML: fetch the model location details \n of registered AI/ML model(s) ML -> rApp: <<R1>> Retrieve AI/ML model information response \n (AI/ML model identifier(s), model location details) end @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 128 Figure 11.5.4.1-1: Retrieve model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.5.5 Required data | For Retrieve AI/ML model information request, the rApp provides the rAppId, and model identifier(s). For Retrieve AI/ML model information response, the AI/ML workflow functions provide model identifier(s) and model location details. 11.6 AI/ML workflow-related use case 6: AI/ML model deployment |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.1 Overview | This use case defines how an rApp requests the deployment of an AI/ML model as an AI/ML model Consumer. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.2 Background and goal of the use case | The AI/ML model deployment procedures are defined as part of the AI/ML workflow services as part of AI/ML workflow services in R1GAP [1]. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.3 Entities/resources involved in the use case | 1) AI/ML workflow functions: a) support model management and exposure functionality to allow an rApp to request the deployment of an AI/ML model. 2) rApp: a) initiates the procedure to deploy the updated artifact version of an AI/ML model being consumed by the rApp. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 129 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.4.1 Retrieve model | Table 11.6.4.1-1: Model deployment Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp requests to deploy an AI/ML model being consumed by the rApp. Actors and Roles - rApp in the role of Model deployment service Consumer. - AI/ML workflow functions in the role of Model deployment service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the Model deployment service. - The rApp is authorized to access the AI/ML model referenced in the deployment request. - The AI/ML model referenced in the deployment request is being consumed by the rApp requesting the deployment. Begins when The rApp determines the need to deployment an AI/ML model. Step 1 (M) The rApp requests the AI/ML workflow functions to deploy an AI/ML model by providing the rAppId, the AI/ML model identifier and the target deployment location of the AI/ML model. Step 2 (M) The AI/ML workflow functions authorize the rApps deployment request. Step 3 (M) The AI/ML workflow functions send the model deployment response to the rApp. The AI/ML workflow functions trigger the rApp deployment procedure. See note. Step 4 (M) The AI/ML workflow functions notify the rApp the model deployment status. Ends when n/a Exceptions n/a Post Conditions The model deployment status is known to the rApp. Traceability REQ-R1-AIML-Deploymodel-FUN1. NOTE: The rApp deployment procedure is not specified in the present document. @startuml!pragma teoz trueskinparam ParticipantPadding 5skinparam BoxPadding 10skinparam defaultFontSize 12skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparentskinparam SequenceGroupBodyBackgroundColor Transparentautonumber box "Non-RT RIC" #whitesmoke box #ivory participant "Model Consumer rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endboxendbox group Model deploymentrApp -> ML : <<R1>> Model deployment request (rAppId, AI/ML Model Identifier,\n the target deployment location of the AI/ML model)ML -> ML: AuthZnote right Check authorization in collaboration with SME functionsend note ML -> rApp : <<R1>> Model deployment response note over ML The rApp deployment procedure is not specified in this specification end note ML -> rApp : <<R1>> Model deployment status notification (AI/ML model identifier, deployment status) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 130 Figure 11.6.4.1-1: Model deployment use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.6.5 Required data | For a model deployment request, the rApp provides the rAppId, AI/ML model identifier, and the target deployment location of the AI/ML model. For a model deployment status notification, AI/ML workflow functions provide AI/ML model identifier and the deployment status. 11.7 AI/ML workflow-related use case 7: AI/ML model performance monitoring - AIML workflow functions consuming performance monitoring service 11.7.1 Overview This use case enables the AI/ML workflow functions as AI/ML model performance monitoring service consumer to subscribe to AI/ML model performance and receive notifications of AI/ML model performance information. 11.7.2 Background and goal of the use case An AI/ML model performance monitoring service consumer can subscribe to AI/ML model performance and receive notifications of AI/ML model performance information. 11.7.3 Entities/resources involved in the use case 1) rApp as AI/ML model performance monitoring service Producer: a) supports the functionality to allow an authorized consumer to subscribe to, and receive notifications of, AI/ML model performance information. 2) AI/ML workflow functions as AI/ML model performance monitoring service Consumer: a) support the functionality to initiate the procedure to subscribe to, and receive notifications of, AI/ML model performance information. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 131 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.7.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.7.4.1 Subscribe AI/ML model performance | Table 11.7.4.1-1: Subscribe AI/ML model performance Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML workflow functions subscribe to AI/ML model performance. Actors and Roles - rApp in the role of AI/ML model performance monitoring service Producer. - AI/ML workflow functions in the role of AI/ML model performance monitoring service Consumer. Assumptions - The supported AI/ML model performance information are part of the AI/ML model registration information. Preconditions - The rApp has deployed a registered AI/ML model. Begins when The AI/ML workflow functions determine the need to monitor the AI/ML model performance of a deployed AI/ML model. Step 1 (M) The AI/ML workflow functions subscribe to AI/ML model performance with AI/ML model identifier, required AI/ML model performance information (optional), periodicity (optional), event notification conditions (optional), and notification URI as parameters. Step 2 (M) The rApp validates the subscription request. Step 3 (M) The rApp creates the subscription. Step 4 (M) The rApp responds to the request with subscription ID as a parameter. Ends when The subscription is created, and AI/ML workflow functions have the subscription ID. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-Performancer-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp end box box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "AI/ML Workflow functions" as AIML end box AIML -> rApp : <<R1>> Subscribe AI/ML model performance request (AI/ML model identifier, \n AI/ML model performance information (optional), Periodicity (optional),\n EventNotificationConditions (optional), Notification URI) rApp --> rApp : Validate request rApp --> rApp : Create subscription rApp -> AIML: <<R1>> Subscribe AI/ML model performance response (Subscription ID) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 132 Figure 11.7.4.1-1: Subscribe AI/ML model performance use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 133 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.7.4.2 Notify AI/ML model performance | Table 11.7.4.2-1: Notify AI/ML model performance Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp notifies subscribed consumers of AI/ML model performance information. Actors and Roles - rApp in the role of AI/ML model performance monitoring service Producer. - AI/ML workflow functions in the role of AI/ML model performance monitoring service Consumer. Assumptions - The supported AI/ML model performance information are part of the AI/ML model registration information. Preconditions - The rApp has deployed a registered AI/ML model. - The AI/ML workflow functions subscribed to AI/ML model performance. Begins when The rApp determines the need to notify subscribed consumers of the performance information of a deployed AI/ML model. Alternative procedure In the case of event-based notifications, the rApp notifies subscribed consumers of the AI/ML model performance information every time an event is triggered in Step 1. Step 1 (M) The rApp checks the occurrence of the notification events specified in the subscription request. Step 2 (M) The rApp notifies the subscribed consumers of the AI/ML model performance information providing the requested AI/ML model performance information and the event triggered as parameters. If no AI/ML model performance information is specified in the subscription request, all supported AI/ML model performance information is provided. Alternative procedure In the case of periodic notifications, the rApp notifies subscribed consumers of the AI/ML model performance information as per the notification periodicity specified in the subscription. Step 3 (M) The rApp notifies the subscribed consumers of the AI/ML model performance information providing the requested AI/ML model performance information. If no AI/ML model performance information is specified in the subscription request, all supported AI/ML model performance information is provided. Ends when The rApp is able to terminate the subscription or the AI/ML model is no longer deployed. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-Performancer-FUN2. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML model performance monitoring \n service Producer rApp" as rApp end box box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "AI/ML Workflow functions" as AIML end box Loop Periodicity of notifications specified in the subscription request Alt Event based notifications rApp --> rApp : Check if specified event is triggered rApp -> AIML : <<R1>> Notify AI/ML model performance (AI/ML model identifier, \n AI/ML model performance information, EventNotificationType) else Periodical notifications rApp -> AIML : <<R1>> Notify AI/ML model performance (AI/ML model identifier, \n AI/ML model performance information) end end ETSI ETSI TS 104 230 V10.0.0 (2026-02) 134 @enduml Figure 11.7.4.2-1: Notify AI/ML model performance use case flow diagram 11.7.5 Required data For the Subscribe AI/ML model performance request, the AI/ML workflow functions provide the AI/ML model identifier and notification URI. In the Subscribe AI/ML model performance request, the AI/ML workflow functions optionally provide the required AI/ML model performance information, Notification event conditions upon which the Notify AI/ML model performance is triggered, and periodicity of the notification. For the Subscribe AI/ML model performance response, the rApp provides the Subscription ID. For the Notify AI/ML model performance, the rApp provides the AI/ML model performance information specified in the Subscribe AI/ML model performance request, as well as the event that triggered the notification if notification event conditions are specified in the Subscribe AI/ML model performance request. If no required AI/ML performance information is specified in the Subscribe AI/ML model performance request, all supported AI/ML model performance information is provided in the Notify AI/ML model performance. 11.8 AI/ML workflow-related use case 8: AI/ML model performance monitoring - rApp consuming performance monitoring service |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.1 Overview | This use case defines how an rApp as AI/ML model performance monitoring service Consumer subscribes to AI/ML model performance and receives notifications of AI/ML model performance information. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.2 Background and goal of the use case | An AI/ML model performance monitoring service Consumer can subscribe to AI/ML model performance and receive notifications of AI/ML model performance information. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 135 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.3 Entities/resources involved in the use case | 1) rApp as AI/ML model performance monitoring service Producer: a) supports the functionality to allow an authorized consumer to subscribe to, and receive notifications of, AI/ML model performance information. 2) rApp as AI/ML model performance monitoring service Consumer: a) supports the functionality to initiate the procedure to subscribe to, and receive notifications of, AI/ML model performance information. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.4.1 Subscribe AI/ML model performance | Table 11.8.4.1-1: Subscribe AI/ML model performance Use Case Stage Evolution / Specification <<Uses>> Related use Goal rApp subscribes to AI/ML model performance. Actors and Roles - rApp in the role of AI/ML model performance monitoring service Producer. - rApp in the role of AI/ML model performance monitoring service Consumer. Assumptions - The supported AI/ML model performance information are part of the AI/ML model registration information. Preconditions - The AI/ML model performance monitoring service producer rApp has deployed a registered AI/ML model. Begins when The AI/ML model performance monitoring service Consumer rApp determines the need to monitor the AI/ML model performance of a deployed AI/ML model. Step 1 (M) The AI/ML model performance monitoring service Consumer rApp subscribe to AI/ML model performance with AI/ML model identifier, required AI/ML model performance information (optional), periodicity (optional), event notification conditions (optional), and notification URI as parameters. Step 2 (M) The AI/ML model performance monitoring service Producer rApp checks with SME functions whether the AI/ML model performance monitoring service Consumer is authorized to subscribe to AI/ML model performance. Step 3 (M) The AI/ML model performance monitoring service Producer rApp validates the subscription request. Step 4 (M) The rApp creates the subscription. Step 5 (M) The rApp responds to the request with subscription ID as a parameter. Ends when The subscription is created, and The AI/ML model performance monitoring service Consumer rApp has the subscription ID. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-Performancer-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "The AI/ML model performance monitoring \n service Producer rApp" as rApp participant " The AI/ML model performance monitoring \n service Consumer rApp " as AIML end box box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant "SME functions" as SME ETSI ETSI TS 104 230 V10.0.0 (2026-02) 136 end box AIML -> rApp : Subscribe AI/ML model performance request (AI/ML model identifier, \n AI/ML model performance information (optional), Periodicity (optional),\n EventNotificationConditions (optional), Notification URI) rApp --> SME : <<R1>> AuthZ note right Check authorization in collaboration with SME functions end note rApp --> rApp : Validate request rApp --> rApp : Create subscription rApp -> AIML: Subscribe AI/ML model performance response (Subscription ID) @enduml Figure 11.8.4.1-1: Subscribe AI/ML model performance use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 137 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.4.2 Notify AI/ML model performance | Table 11.8.4.2-1: Notify AI/ML model performance Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp notifies subscribed consumers of AI/ML model performance information. Actors and Roles - rApp in the role of AI/ML model performance monitoring service Producer. - rApp in the role of AI/ML model performance monitoring service Consumer. Assumptions - The supported AI/ML model performance information are part of the AI/ML model registration information. Preconditions - The rApp has deployed a registered AI/ML model. - The AI/ML model performance monitoring service consumer rApp subscribed to AI/ML model performance. Begins when The rApp determine the need to notify subscribed consumers of the performance information of a deployed AI/ML model. Alternative procedure In the case of event-based notifications, the rApp notifies subscribed consumers of the AI/ML model performance information every time an event is triggered in Step 1. Step 1 (M) The rApp checks the occurrence of the notification events specified in the subscription request. Step 2 (M) The rApp notifies the subscribed consumers of the AI/ML model performance information providing the requested AI/ML model performance information and the event triggered as parameters. If no AI/ML model performance information is specified in the subscription request, all supported AI/ML model performance information is provided. Alternative procedure In the case of periodic notifications, the rApp notifies subscribed consumers of the AI/ML model performance information as per the notification periodicity specified in the subscription. Step 3 (M) The rApp notifies the subscribed consumers of the AI/ML model performance information providing the requested AI/ML model performance information. If no AI/ML model performance information is specified in the subscription request, all supported AI/ML model performance information is provided. Ends when The rApp is able to terminate the subscription or the AI/ML model is no longer deployed. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-Performancer-FUN2. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "AI/ML model performance monitoring \n service Producer rApp" as rApp participant " The AI/ML model performance monitoring \n service Consumer rApp " as AIML end box Loop Periodicity of notifications specified in the subscription request Alt Event based notifications rApp --> rApp : Check if specified event is triggered rApp -> AIML : Notify AI/ML model performance (AI/ML model identifier, \n AI/ML model performance information,EventNotificationType) else Periodical notifications rApp -> AIML : Notify AI/ML model performance (AI/ML model identifier, \n AI/ML model performance information) end end @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 138 Figure 11.8.4.2-1: Notify AI/ML model performance use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.8.5 Required data | For the Subscribe AI/ML model performance request, the AI/ML model performance monitoring service Consumer rApp provides the AI/ML model identifier and notification URI. In the Subscribe AI/ML model performance request, the AI/ML model performance monitoring service Consumer rApp optionally provides the required AI/ML model performance information, Notification event conditions upon which the Notify AI/ML model performance is triggered, and periodicity of the notification. For the Subscribe AI/ML model performance response, the AI/ML model performance monitoring service Producer rApp provides the Subscription ID. For the Notify AI/ML model performance, the AI/ML model performance monitoring service producer rApp provides the AI/ML model performance information specified in the Subscribe AI/ML model performance request, as well as the event that triggered the notification if notification event conditions are specified in the Subscribe AI/ML model performance request. If no required AI/ML performance information is specified in the Subscribe AI/ML model performance request, all supported AI/ML model performance information is provided in the Notify AI/ML model performance. 11.9 AI/ML workflow-related use case 9: AI/ML model training capability registration and deregistration |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.1 Overview | This use case defines how an rApp registers and deregisters AI/ML model training capability. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.2 Background and goal of the use case | The register AI/ML model training capability and deregister AI/ML model training capability are optional procedures defined as part of the AI/ML model training capability registration service in R1GAP [1]. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 139 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.3 Entities/resources involved in the use case | 1) AI/ML workflow functions: a) support functionality to allow an rApp to register and deregister AI/ML model training capability; b) support validation of AI/ML model training capability information. 2) rApp: a) initiates the procedure to register and deregister AI/ML model training capability. |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.4 Solutions | |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.4.1 Register AI/ML model training capability | Table 11.9.4.1-1: AI/ML Model training capability registration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp registers an AI/ML model training capability. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services. - The rApp is registered as producer of AI/ML model training with SME, it intends to register its training capability in AI/ML workflow functions. Begins when The rApp determines the need to register an AI/ML model training capability. Step 1 (M) The rApp requests the AI/ML workflow functions to register an AI/ML model training capability by providing the rAppId and AI/ML model training capability information. Step 2 (M) The AI/ML workflow functions check with SME functions whether the rApp is registered as producer of AI/ML model training service. Step 3 (M) The AI/ML workflow functions register the rApp's AI/ML model training capability. Step 4 (M) The AI/ML workflow functions respond to rApp with successful AI/ML model training capability registration along with AI/ML model training capability registration ID. Ends when The rApp was able to register an AI/ML model training capability. Exceptions n/a Post Conditions The AI/ML model training capability is registered. The rApp can query, update or delete the AI/ML model training capability registration. Traceability REQ-R1-AI/ML-Registertraincap-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox box #ivory participant "Training service Producer rApp" as src ETSI ETSI TS 104 230 V10.0.0 (2026-02) 140 endbox endbox src -> ML: <<R1>> Register AI/ML model training capability request (rAppId, AI/ML model training capability information) ML -> ML: Authz ML -> ML: register AI/ML model training capability ML -> src: <<R1>> Register AI/ML model training capability response (AI/ML model training capability registration ID) @enduml Figure 11.9.4.1-1: Register AI/ML model training capability use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.4.2 Deregister AI/ML model training capability | Table 11.9.4.2-1: AI/ML Model training capability deregistration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp deregisters an AI/ML model training capability. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions The rApp is not producing any AI/ML model training capability. Preconditions The rApp has registered an AI/ML model training capability. The rApp is authorized to access the AI/ML model management and exposure services. Begins when The rApp determines the need to deregister an AI/ML model training capability that it no longer intends to produce. Step 1 (M) The rApp requests AI/ML workflow functions to deregister an AI/ML model training capability by providing rAppId and AI/ML model training capability registration ID. Step 2 (M) The AI/ML workflow functions remove the registration of the rApp as producer of AI/ML model training capability. Step 3 (M) The AI/ML workflow functions respond to the rApp with successful deregistration. Ends when The rApp was able to deregister the AI/ML model training capability. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AI/ML-Registertraincap-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML ETSI ETSI TS 104 230 V10.0.0 (2026-02) 141 endbox box #ivory participant "Training service Producer rApp" as src endbox endbox src -> ML: <<R1>> Deregister AI/ML model training capability (rAppId, AI/ML model training capability registration ID) ML -> ML: Deregister AI/ML model training capability ML -> src: <<R1>> Deregister AI/ML model training capability response @enduml Figure 11.9.4.2-1: Deregister AI/ML model training capability use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.4.3 Update AI/ML model training capability registration | Table 11.9.4.3-1: Update AI/ML model training capability registration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp updates the registration of an AI/ML model training capability. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services for updating an AI/ML model training capability registration. - The rApp has registered an AI/ML model training capability. Begins when The AI/ML model training service producer rApp determines the need to update the registration of an AI/ML model training capability. Step 1 (M) The AI/ML model training service producer rApp requests to update the registration of an AI/ML model training capability by providing rAppId, AI/ML model training capability registration ID, updated model training capability registration information, etc. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to update the registration. Step 3 (M) The AI/ML workflow functions validate the update AI/ML model training capability registration request. Step 4 (M) The AI/ML workflow functions update the registration of the AI/ML model training capability. Step 5 (M) The AI/ML workflow functions respond to the rApp with successful update AI/ML model training capability registration response. Ends when The rApp was able to update the registration of the AI/ML model training capability. Exceptions n/a Post Conditions The rApp can query, update or delete the AI/ML model training capability registration. Traceability REQ-R1-AI/ML-Registertraincap-FUN1. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 142 @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as app endbox box "Non-anchored functions \n in SMO/Non-RT RIC FWK" #cadetBlue participant "AI/ML workflow functions" as aif endbox endbox autonumber app -> aif: <<R1>> Update model training capability registration request \n (rAppId, AI/ML model training capability registration identifier, \n updated model training capability registration information) activate aif aif --> aif: AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Validate aif --> aif: Update model training capability\n registration information app <- aif: <<R1>> Update model training capability registration response deactivate aif @enduml Figure 11.9.4.3-1: Update registration of an AI/ML training capability model use case flow diagram ETSI ETSI TS 104 230 V10.0.0 (2026-02) 143 |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.4.4 Query AI/ML model training capability registration | Table 11.9.4.4-1: Query AI/ML model training capability registration Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp queries an AI/ML model training capability registration. Actors and Roles - rApp in the role of AI/ML model management and exposure services Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure services Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure services for registering an AI/ML model. - The AI/ML model training capability has been registered by the rApp. Begins when The AI/ML model training service producer rApp determines the need to query the registration of a registered AI/ML model training capability. Step 1 (M) The AI/ML model training service producer rApp requests to query the registration of an AI/ML model training capability by providing rAppId and AI/ML model training capability registration ID. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to query the AI/ML model training capability registration. Step 3 (M) The AI/ML workflow functions look up the information of queried AI/ML model training capability. Step 4 (M) The AI/ML workflow functions respond to rApp with AI/ML model training capability information. Ends when The AI/ML model training service producer rApp was able to query the AI/ML model training capability registration. Exceptions n/a Post Conditions The rApp can query, update, or delete the AI/ML model training capability registration. Traceability REQ-R1-AI/ML-Registertraincap-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as src endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox src -> ML: <<R1>> Query AI/ML model training capability request (rAppId, AI/ML model training capability registration ID) ML -> ML: Authz note right Check authorization in collaboration with SME functions end note ML -> ML: Look up AI/ML model training capability registration ML -> src: <<R1>> Query AI/ML model training capability response (AI/ML model training capability information) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 144 Figure 11.9.4.4-1: Update registration of an AI/ML training capability model use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.9.5 Required data | The AI/ML model training capability information includes but not limited to supported training frameworks, supported library packages, supported library versions, resource information. For registering an AI/ML model training capability, the rApp provides the rAppId and AI/ML model training capability information. On successful registration, AI/ML workflow functions provide an AI/ML model training capability registration ID for the registration of the AI/ML model training capability information. For deregistering an AI/ML model training capability, the rApp provides the rAppId and the AI/ML model training capability registration ID. For updating the registration of an AI/ML model training capability, the rApp needs to provide the rAppId, the AI/ML model training capability registration ID and the updated AI/ML model training capability information or the modified part of the AI/ML model training capability information. For querying the registration of an AI/ML model training capability, the rApp needs to provide the rAppId and the AI/ML model training capability registration ID. 11.10 AI/ML workflow-related use case 10: AI/ML model inference - AI/ML workflow functions producing AI/ML inference 11.10.1 Overview This use case enables an rApp to request and cancel an inference for an AI/ML model. 11.10.2 Background and goal of the use case The request and cancel AI/ML model inference procedures are defined as part of the AI/ML workflow services in R1GAP [1]. 11.10.3 Entities/resources involved in the use case 1) AI/ML workflow functions in the role of AI/ML model inference service Producer: a) support functionality allowing rApps to request, and cancel the inference of a registered AI/ML Model. 2) rApp in the role of AI/ML model inference service Consumer: a) initiates the procedure to request and cancel the inference of an AI/ML model. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 145 11.10.4 Solutions |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.10.4.1 Request AI/ML model inference | Table 11.10.4.1-1: Request AI/ML model inference use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML inference services Consumer requests inference for an AI/ML model. Actors and Roles - AI/ML inference services Consumer in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML inference services Consumer is deployed, authenticated, and authorized to consume AI/ML inference services. Begins when The AI/ML inference services Consumer determines the need to initiate the inference an AI/ML model. Step 1 (M) The AI/ML inference services Consumer requests the AI/ML workflow functions to infer an AI/ML model providing rAppId and model identifier. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML inference services Consumer is authorized to request inference for an AI/ML model. Step 3 (M) The AI/ML workflow functions validate the request. Step 4 (M) The AI/ML workflow functions respond to the AI/ML inference services Consumer with inference job identifier as a parameter. Ends when The AI/ML inference services Consumer is able to obtain the inference job identifier. Exceptions n/a Post Conditions The inference job exists, and the inference service Consumer can cancel the AI/ML model inference. Traceability REQ-R1-AIML-inference-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox rApp -> ML: <<R1>> Request AI/ML model inference request \n (rAppId, AI/ML model identifier) ML -> ML: AuthZ note right Check authorization in Collaboration with SME functions end note ML-> ML :validate ML -> rApp: <<R1>> Request AI/ML model inference response \n (inference job identifier) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 146 Figure 11.10.4.1-1: Request AI/ML model inference use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.10.4.2 Cancel AI/ML model inference | Table 11.10.4.2-1: Cancel AI/ML model inference use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML inference services Consumer cancels the AI/ML model inference that it has previously requested. Actors and Roles - AI/ML inference services Consumer in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions n/a Preconditions - The AI/ML inference services Consumer is deployed, authenticated, and authorized to consume AI/ML inference services. - The inference job exists, and the AI/ML inference Consumer is aware of the job identifier. Begins when The AI/ML inference services Consumer determines the need to cancel the inference of an AI/ML model. Step 1 (M) The AI/ML inference services Consumer requests the AI/ML workflow functions to cancel the inference for an AI/ML model providing rAppId and inference job identifier. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML inference services Consumer is authorized to cancel the inference job for an AI/ML model. Step 3 (M) The AI/ML workflow functions cancel the inference job. Step 4 (M) The AI/ML workflow functions respond to the AI/ML inference services Consumer rApp with cancellation of inference job. Ends when The AI/ML inference service job has been cancelled. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-inference-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber ETSI ETSI TS 104 230 V10.0.0 (2026-02) 147 box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox rApp -> ML: <<R1>> Cancel AI/ML model inference request \n (rAppId,inference job identifier) ML -> ML: AuthZ note right Check authorization in Collaboration with SME functions end note ML -> ML : Stop inference job ML -> rApp: <<R1>> Cancel AI/ML model inference response @enduml Figure 11.10.4.2-1: Cancel AI/ML model inference use case flow diagram 11.10.5 Required data For creating AI/ML model inference job, the AI/ML inference services Consumer needs to provide its rAppId and the model identifier. The AI/ML workflow functions respond with the creation of inference job by providing the inference job identifier. For cancellation of an AI/ML inference job, the AI/ML inference services Consumer needs to provide its rAppId and the inference job identifier. 11.11 AI/ML workflow-related use case 11: AI/ML model change subscription 11.11.1 Overview This use case defines how an rApp subscribes to registered AI/ML models. It enables an rApp to subscribe to and unsubscribe from notifications regarding changes in the registered AI/ML models. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 148 11.11.2 Background and goal of the use case The subscribe AI/ML models changes procedure, unsubscribe AI/ML models changes procedure and notify AI/ML models changes procedure are defined as part of the AI/ML workflow services in R1GAP [1]. 11.11.3 Entities/resources involved in the use case 1) AI/ML workflow functions: a) support functionality to allow an rApp to subscribe and unsubscribe the registered AI/ML model changes; b) support functionality to notify rApp about changes of registered AI/ML model; c) support validation of selection criteria. 2) rApp: a) initiates the procedure to subscribe and unsubscribe registered AI/ML model changes; b) support functionality to receive notification regarding changes of subscribed AI/ML model. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 149 11.11.4 Solutions |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.11.4.1 Subscribe AI/ML model changes | Table 11.11.4.1-1: Subscribe AI/ML model changes Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp subscribes to notifications regarding changes of AI/ML models. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure service. - The rApp is aware of the AI/ML model identifier of the AI/ML model. Begins when The rApp determines the need to subscribe to notifications regarding changes of registered AI/ML models. Step 1 (M) The rApp requests the AI/ML workflow functions to subscribe to notifications regarding AI/ML model changes with rAppId and AI/ML model identifier selection criteria. Step 2 (M) The AI/ML workflow functions receive the AI/ML model subscription request, and check whether the rApp is authorized to subscribe to identified AI/ML models. Step 3 (M) The AI/ML workflow functions validate the subscribe AI/ML model request. Step 4 (M) The AI/ML workflow functions establish the AI/ML model subscription. Step 5 (M) The AI/ML workflow functions send back a response with a subscription identifier matching the request in Step 1 to rApp. Ends when The rApp was able to subscribe to notifications regarding AI/ML model changes. Exceptions n/a Post Conditions - The rApp can receive notifications when changes are made to any of the AI/ML models that matches AI/ML model identifier selection criteria. - The rApp can unsubscribe from the notifications. Traceability n/a NOTE: AI/ML model identifier includes model name, model version and artifact version. AI/ML model identifier selection criteria could be upper and/or lower bounds of model version and artifact version number, or list of individual model version and artifact version numbers, etc. Selection criteria is used to identify AI/ML models to be subscribed, which can be applied to part, or all of AI/ML model identifier components. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox rApp -> ML: <<R1>> Subscribe AI/ML model change request (rAppId, \n AI/ML model identifier selection criteria) ML -> ML: Authz note right Check authorization in collaboration with SME functions ETSI ETSI TS 104 230 V10.0.0 (2026-02) 150 end note ML -> ML: Validate ML -> ML: Create subscription ML -> rApp: <<R1>> Subscribe AI/ML model change response (subscription identifier) @enduml Figure 11.11.4.1-1: Subscribe AI/ML model changes use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.11.4.2 Notify AI/ML model changes | Table 11.11.4.2-1: Notify AI/ML model changes Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp receives a notification regarding a change of a registered AI/ML model. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure service. - The rApp is subscribed to notifications. Begins when The AI/ML workflow functions determine to send a notification regarding the change of a subscribed AI/ML model to the subscribing rApp. Step 1 (M) The AI/ML workflow functions send a notification to the subscribing rApp with AI/ML model identifier and available change details. Ends when The rApp was able to receive the notification. Exceptions n/a Post Conditions n/a Traceability n/a @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox ETSI ETSI TS 104 230 V10.0.0 (2026-02) 151 box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox ML -> rApp: <<R1>> Notify AI/ML model changes (AI/ML model identifier, available change details) @enduml Figure 11.11.4.2-1: Notify AI/ML model changes use case flow diagram |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.11.4.3 Unsubscribe AI/ML model changes | Table 11.11.4.3-1: Unsubscribe AI/ML model changes Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp unsubscribes from notifications. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions - The rApp is authorized to access the AI/ML model management and exposure service. - The rApp is subscribed to notifications regarding AI/ML model changes. Begins when The rApp determines the need to unsubscribe from notifications regarding changes of registered AI/ML model. Step 1 (M) The rApp requests the AI/ML workflow functions to unsubscribe from notifications regarding AI/ML model changes by providing the rAppId and subscription identifier. Step 2 (M) The AI/ML workflow functions check whether the rApp is authorized to unsubscribe from notifications regarding AI/ML model changes. Step 3 (M) The AI/ML workflow functions cancel the subscription. Step 4 (M) The AI/ML workflow functions respond to the request. Ends when The rApp was able to unsubscribe from notifications regarding changes of registered AI/ML model. Exceptions n/a Post Conditions n/a Traceability n/a @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox ETSI ETSI TS 104 230 V10.0.0 (2026-02) 152 rApp -> ML: <<R1>> Unsubscribe AI/ML model change request (rAppId, subscription identifier) ML -> ML: Authz note right Check authorization in collaboration with SME functions end note ML -> ML: Cancel subscription ML -> rApp: <<R1>> Unsubscribe AI/ML model change response @enduml Figure 11.11.4.3-1: Unsubscribe AI/ML model changes use case flow diagram 11.11.5 Required data For subscription to notifications regarding changes of registered AI/ML model, the rApp provides the rAppId and the AI/ML model identifier selection criteria. The AI/ML workflow functions send back a subscription identifier in response. The AI/ML workflow functions send notifications to the subscribing rApp by providing the AI/ML model identifier(s) and available change details of AI/ML model(s). For unsubscribing from notifications regarding changes of registered AI/ML model, the rApp needs to send the rAppId and AI/ML model subscription identifier. 11.12 AI/ML workflow-related use case 12: Query of AI/ML Model inference capabilities - AI/ML workflow functions producing AI/ML inference 11.12.1 Overview This use case enables an rApp to query the inference capability information of an AI/ML model. 11.12.2 Background and goal of the use case The query AI/ML model inference capabilities procedure are defined as part of the AI/ML workflow services in R1GAP [1]. 11.12.3 Entities/resources involved in the use case 1) AI/ML workflow functions in the role of AI/ML model inference service Producer: a) Support functionality allowing rApps to query the inference capability of an AI/ML Model. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 153 2) rApp in the role of AI/ML model inference service Consumer: a) Initiates the procedure to query the inference capability of an AI/ML model. 11.12.4 Solutions |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.12.4.1 Query AI/ML model inference capability | Table 11.12.4.1-1: Query AI/ML model inference capability use case Use Case Stage Evolution / Specification <<Uses>> Related use Goal The AI/ML inference services Consumer retrieves the AI/ML model inference capability information. Actors and Roles - AI/ML inference services Consumer in the role of Service Consumer. - The AI/ML workflow functions in the role of Service Producer. Assumptions Service Consumer is authorized to request inference capability information of a registered AI/ML model. Preconditions - The AI/ML inference services Consumer is deployed, authenticated, and authorized to consume AI/ML inference services. Begins when The AI/ML inference services Consumer determines the need to query the inference capability information of a registered AI/ML model. Step 1 (M) The AI/ML inference services Consumer requests the AI/ML workflow functions to provide inference capability information by passing on the query parameters such as model identifier and associated model information. Step 2 (M) The AI/ML workflow functions check with SME functions whether the AI/ML inference services Consumer is authorized to request the inference capability information for a registered AI/ML model. Step 3 (M) The AI/ML workflow functions respond to the AI/ML inference services Consumer rApp with the requested inference capability information that matches the query criteria. Ends when The AI/ML model inference capability information has been retrieved. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AIML-inference-FUN2. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid skinparam SequenceGroupBackgroundColor Transparent skinparam SequenceGroupBodyBackgroundColor Transparent autonumber box "Non-RT RIC" #whitesmoke box #ivory participant "rApp" as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework " #cadetBlue participant "AI/ML workflow functions" as ML endbox endbox rApp -> ML: <<R1>> Query AI/ML model inference capability request \n (rAppId, query criteria) ML -> ML: AuthZ note right Check authorization in Collaboration with SME functions end note ML -> rApp: <<R1>> Query AI/ML model inference capability response \n (inference capability information) @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 154 Figure 11.12.4.1-1: Query AI/ML model inference capability use case flow diagram 11.12.5 Required data For querying of an AI/ML model inference capability information, the AI/ML inference services Consumer needs to provide its rAppId and the query criteria (including the model identifier and model information). The AI/ML workflow functions respond with inference capability information such as model identifier or inference latency. 11.13 AI/ML Workflow-related use case 13: Query of AI/ML Model training capability 11.13.1 Overview This use case defines how an rApp queries AI/ML model training capability. 11.13.2 Background and goal of the use case The query AI/ML model training capability is an optional procedure defined as part of the AI/ML model training capability query service in R1GAP [1]. 11.13.3 Entities/resources involved in the use case 1) AI/ML workflow functions in the role of AI/ML model management and exposure service producer: a) Supports functionality allowing rApps to query the registered AI/ML model training capability; 2) rApp in the role of AI/ML model management and exposure service consumer: a) Initiates the procedure to query the registered AI/ML model training capability. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 155 11.13.4 Solutions |
e341f05abb94e155b675f29c6779693c | 104 230 | 11.13.4.1 Query AI/ML model training capability | Table 11.13.4.1-1: Query AI/ML model training capability. Use Case Stage Evolution / Specification <<Uses>> Related use Goal The rApp queries registered AI/ML model training capability. Actors and Roles - rApp in the role of AI/ML model management and exposure service Consumer. - AI/ML workflow functions in the role of AI/ML model management and exposure service Producer. Assumptions n/a Preconditions The rApp is authorized to access the AI/ML workflow services. At least one AI/ML model training capability is registered with the AI/ML workflow functions. Begins when The rApp determines the need to query the registered AI/ML model training capability. Step 1 (M) The rApp requests the AI/ML workflow functions for the information on the AI/ML model training capability by providing rAppId and selection criteria. Step 2 (M) The AI/ML workflow functions check if the rApp is authorized to query the registered AI/ML model training capability. Step 3 (M) The AI/ML workflow functions look up the information of queried AI/ML model training capability. Step 4 (M) The AI/ML workflow functions provide the information about the registered AI/ML model training capability. Ends when The AI/ML model training capability has been received. Exceptions n/a Post Conditions n/a Traceability REQ-R1-AI/ML-Registertraincap-FUN1. @startuml !pragma teoz true skinparam ParticipantPadding 5 skinparam BoxPadding 10 skinparam defaultFontSize 12 skinparam lifelineStrategy solid autonumber box "Non-RT RIC" #whitesmoke box #ivory participant rApp as rApp endbox box " Non-anchored functions in SMO/Non-RT RIC Framework" #cadetBlue participant " AI/ML workflow functions " as aif endbox rApp ->aif:<<R1>> Query AI/ML model training capability request\n(rAppId, Selection criteria) activate aif aif --> aif:AuthZ note right Check authorization in collaboration with SME functions end note aif --> aif: Look up AI/ML model training capability registration aif -> rApp :<<R1>> Query AI/ML model training capability response\n(AI/ML model Trainer rAppIds, AI/ML model training capability information) deactivate aif @enduml ETSI ETSI TS 104 230 V10.0.0 (2026-02) 156 Figure 11.13.4.1-1: Query AI/ML model training capability use case flow diagram 11.13.5 Required data For the Query AI/ML model training capability request, the rApp provides the rAppId and selection criteria. The AI/ML workflow functions respond with information which includes the AI/ML model Trainer rAppIds and AI/ML model training capability information that match the filtering criteria. The AI/ML model training capability information contains training platform information and training resource information. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 157 Annex A (informative): Change history Date Version Information about changes 13.03.2025 V10.00 Added use cases for AI/ML model training capability registration and data job status. 21.11.2024 V09.00 Added use cases for AIML inference capability information Query. 18.07.2024 V08.00 Added use cases for AI/ML model training capability registration, Query for data subscription, AI/ML model inference use case for create and delete and updated the SME subscription use case. 20.04.2024 V07.00 Added the use cases for AI/ML performance monitoring use cases and updated the document with Producer and Consumer roles. 20.11.2023 V06.00 Added the use cases and requirements to Data management and exposure services, A1 related services, AI/ML model workflow services, updated the data type to DME type. 29.07.2023 V05.00 Published as Final version 05.00. 24.03.2023 V04.00 Published as Final Version 04.00. 10.11.2022 V03.00 Published as Final Version 03.00. 29.07.2022 V02.00 Published as Final version 02.00. 01.04.2022 V01.00 Published as Final version 01.00. ETSI ETSI TS 104 230 V10.0.0 (2026-02) 158 History Version Date Status V10.0.0 February 2026 Publication |
314690db30e0da820145d5cd1eef3bc5 | 104 141 | 1 Scope | The present document has the goal of defining practical and effective methods for planning wireless backhaul links according to the novel approach detailed in ETSI GR mWT 028 [i.1]. One key area of investigation and contribution is the design of appropriate prediction models of the aggregated traffic demand statistics of any backhaul link as a function of few high-level and easy-to-know RAN-related behavioural features, such as the average and the peak values of the transported throughput, or the configuration of the various radio access technologies implemented in the connected sites. The derived prediction models are conceived with the goal of being easily embedded within the software framework of the currently available backhaul planning tools, thus pursuing the strategic benefit of making the choice of the link traffic demand distribution required for enabling the New KPIs methodology (namely, for computing the BTA metric) a completely transparent process for the end user. In this context, the present document discloses an analytical procedure targeted at deriving a pessimistic estimation of the BTA of any backhaul link on the basis of the sole knowledge of the average and the peak values of its traffic demand, that could be used for a conservative and practical network planning. Secondly, the present document describes a methodology for conducting measurement campaigns on live transport networks and for processing the collected data in order to create a discrete set of reference throughput demand distributions to be used in the assessment of the BTA of any backhaul link, after a proper identification and classification of its main deployment features (e.g. number and type of transported radio access layers) and statistical properties. |
314690db30e0da820145d5cd1eef3bc5 | 104 141 | 2 References |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.