hash stringlengths 32 32 | doc_id stringlengths 5 12 | section stringlengths 5 1.47k | content stringlengths 0 6.67M |
|---|---|---|---|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5 Split operation event subscription
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5.1 General
| Clause 8.14.2.5.2 and clause 8.14.2.5.3 together illustrate the split operation subscribe/notify model.
Clause 8.14.2.5.4 illustrates the split operation subscription update procedure.
Clause 8.14.2.5.5 illustrates the split operation unsubscribe procedure.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5.2 Subscribe
| Figure 8.14.2.5.2-1 illustrates the procedure for an AIMLE client or VAL server to subscribe with the AIMLE server to be notified of events related to split AI/ML operation.
Pre-conditions:
1. The AIMLE client or VAL server has received information (e.g. URI, IP address) related to the AIMLE Server;
2. The AIMLE client or VAL server has received security credentials authorizing it to communicate with the AIMLE server;
Figure 8.14.2.5.2-1: Split operation subscription
1. The requestor (e.g., AIMLE client or VAL server) sends a split operation subscribe request to the AIMLE server. The request includes information defined in Table 8.14.3.12-1.
a. The requestor (e.g., AIMLE client or VAL server) may include the "split operation pipeline information" event to indicate the AIMLE server to notify the requestor when a split operation profile is created or an existing pipeline is updated (e.g., node is added or removed, stage order or ML models are modified, etc.) or deleted and satisfy the discovery filters of the subscription request.
b. The requestor (e.g., AIMLE client or VAL server) may include the " split operation assistance information" event to indicate the AIMLE server to notify the requestor when the AIMLE server determines assistance information related to the split operation usage.
2. Upon receiving the request from the requestor, the AIMLE server validates if the requestor is authorized for the request. If the requestor is authorized, the AIMLE server creates the subscription and stores the subscription information.
3. The AIMLE server sends a split operation subscribe response to the requestor. If the AIMLE server has created the subscription, the response includes an indication of success, the subscription identity and may include an expiration time; to maintain the subscription, the requestor shall send a subscription update request before the expiration time, otherwise the split operation subscription expires. If the AIMLE server has not created the subscription, the response includes an indication of failure and may include a reason for failure.
If the subscription is for "assistance information", the AIMLE Server considers the assistance information included in the subscription request and can perform the following:
The AIMLE Server aggregates the collected assistance information from NEF, NWDAF, and/or ADAES to generate assistance information, e.g. Time (time point(s) or time window(s)) to deliver the task or data for the split operation. Or the AIMLE server performs inference using ML model to generate assistance information, e.g. achievable QoS with current configuration for task or data delivery, or suggestion of QoS for task or data delivery.
- For assistance information from NEF, the AIMLE Server, acting as AF, sends a request to NEF for assistance information (e.g. PDTQ on recommended time windows for AI/ML operations with QoS, QoS monitoring) as described in clause 4.16.15 of 3GPP TS 23.502 [3]. The request may include the information of the AI/ML task/ML model/data (e.g. size of the task or ML model, or data volume), information of the receiving nodes, requirement on the delivery/distribution (e.g. time budget, time critical or not, QoS), maximum time for complete the distribution/delivery.
- For assistance information from NWDAF, the AIMLE Server can send a request or subscription request to NWDAF for analytics (e.g. E2E data volume transfer time, DN performance, Network performance, UE mobility). The details parameters in the requests to NWDAF for analytics are given in 3GPP TS 23.288 [2].
- For assistance information from ADAES, the AIMLE Server can send a request or subscription request to ADAES for analytics (e.g. slice-specific application performance analytics, UE-to-UE application performance analytics). The details parameters in the requests to ADAES for analytics are given in 3GPP TS 23.436 [4].
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5.3 Notify
| Figure 8.14.2.5.3-1 illustrates the split operation notify operation between the AIMLE server and an AIMLE client or VAL server.
Pre-conditions:
2. The AIMLE client or VAL server has subscribed for split operation with the AIMLE Server;
Figure 8.14.2.5.3-1: Split operation notification
1. The AIMLE server detects event(s) that satisfies trigger conditions for notifying a subscriber (e.g., AIMLE client or VAL server) according to subscribed events.
a. If the subscribed event is for "split operation pipeline information" and the AIMLE server detects that a new split operation profile is created or an existing pipeline is updated (e.g., node is added or removed, stage order or ML models are modified, etc.) or deleted and satisfy the discovery filters of the subscription, the AIMLE server notifies the subscriber (e.g., AIMLE client or VAL server) accordingly;
b. If the subscribed event is for "split operation assistance information" and the AIMLE server has determined assistance information related to split operation usage, the AIMLE server notifies the subscribers (e.g., AIMLE client or VAL server) accordingly.
2. The AIMLE server sends a split operation notification to the requestor indicating the event. The notification includes a subscription identity and may include a split operation profile(s), split operation node information, or split operation assistance information dependent on the associated event.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5.4 Subscription update
| Figure 8.14.2.5.4-1 illustrates the procedure for an AIMLE client or VAL server to update a subscription with the AIMLE server.
Pre-conditions:
2. The AIMLE client or VAL server has subscribed for split operation with the AIMLE Server;
Figure 8.14.2.5.4-1: Split operation subscription update
1. The requestor (e.g., AIMLE client or VAL server) sends a split operation subscription update request to the AIMLE server. The request includes information defined in Table 8.14.3.15-1.
2. Upon receiving the request from the requestor, the AIMLE server validates if the requestor is authorized for the request. If the requestor is authorized, the AIMLE server updates the subscription information.
3. The AIMLE server sends a split operation subscription update response to the requestor. If the AIMLE server has updated the subscription, the response includes an indication of success and may include an expiration time. To maintain the subscription, the requestor shall send a subscription update request before the expiration time, otherwise the split operation subscription expires. If the AIMLE server has not updated the subscription, the response includes an indication of failure and may include a reason for failure.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.5.5 Unsubscribe
| Figure 8.14.2.5.5-1 illustrates the procedure for an AIMLE client or VAL server to unsubscribe with the AIMLE server.
Pre-conditions:
1. The AIMLE client or VAL server has subscribed for split operation with the AIMLE Server;
Figure 8.14.2.5.4-1: Split operation unsubscribe
1. The requestor (e.g., AIMLE client or VAL server) sends a split operation unsubscribe request to the AIMLE server. The request includes information defined in Table 8.14.3.17-1.
2. Upon receiving the request from the requestor, the AIMLE server validates if the requestor is authorized for the request. If the requestor is authorized, the AIMLE server cancels the subscription associated with the subscription identifier.
3. The AIMLE server sends a split operation unsubscribe response to the requestor. If the AIMLE server has canceled the subscription, the response includes an indication of success. If the AIMLE server has not canceled the subscription, the response includes an indication of failure and may include a reason for failure.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.6 Split operation pipeline update
| Figure 8.14.2.6-1 illustrates the procedure for an AIMLE client to update an instance of a split operation pipeline at the AIMLE server. The AIMLE client or VAL client can determine that some of the nodes in the split operation pipeline may not be able to perform the required operation and decides to modify the pipeline either to add or remove the nodes.
Pre-conditions:
1. The AIMLE client has received information related to split operation pipeline profile as specified in clause 8.14.2.3.
Figure 8.14.2.6-1: Split operation pipeline update
1. The AIMLE client sends a split operation pipeline update request to the AIMLE server. The request includes information defined in Table 8.14.3.19-1.
2. Upon receiving the request from the AIMLE client, the AIMLE server validates if the requestor is authorized for the request. If the requestor is authorized, the AIMLE server validates if the requested split operation pipeline can be updated based on the split operation pipeline identifier included in the request.
If the requestor is authorized and a split operation pipeline profile is determined, the AIMLE server updates the split operation pipeline profile and notifies the appropriate processing nodes indicated in the request about their inclusion or exclusion in the split operation pipeline as described in clause 8.14.2.5. The AIMLE server also notifies the already existing nodes about modification of the existing pipeline as described in clause 8.14.2.5.
3. The AIMLE server sends a split operation pipeline update response message to the AIMLE client. If the AIMLE server has updated an instance of a split operation pipeline profile, the response includes an indication of success, and the corresponding split operation profile. Otherwise, the response includes an indication of failure and may include a reason for failure.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.2.7 Split operation pipeline delete
| Figure 8.14.2.7-1 illustrates the procedure for an AIMLE client to delete an instance of a split operation pipeline at the AIMLE server. The AIMLE client or VAL client can determine to delete a split operation pipeline when it is no longer needed.
Pre-conditions:
1. The AIMLE client has received information related to split operation pipeline profile as specified in clause 8.14.2.3.
Figure 8.14.2.7-1: Split operation pipeline delete
1. The AIMLE client sends a split operation pipeline delete request to the AIMLE server. The request includes information defined in Table 8.14.3.21-1.
2. Upon receiving the request from the AIMLE client, the AIMLE server validates if the requestor is authorized for the request. If the requestor is authorized, the AIMLE server validates if the requested split operation pipeline can be deleted based on the split operation pipeline identifier included in the request.
If the requestor is authorized and a split operation pipeline is determined, the AIMLE server deletes the split operation pipeline profile and notifies the appropriate processing nodes about deletion of the split operation pipeline as described in clause 8.14.2.5.
3. The AIMLE server sends a split operation pipeline delete response message to the AIMLE client. If the AIMLE server has deleted a split operation profile, the response includes an indication of success. Otherwise, the response includes an indication of failure and may include a reason for failure.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.1 General
| The following information flows are specified for operation splitting.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.2 Split operation pipeline discovery request
| Table 8.14.3.2-1 shows the request sent by a AIMLE client to the AIMLE server for split operation pipeline discovery request.
Table 8.14.3.2-1: Split operation pipeline discovery request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL client ID, AIMLE client ID, UE identifier)
Security credentials
M
The security credentials of the requestor.
Split operation discovery filters
M
Split operation discovery filters.
> stage information
M
Information about the split operation stages (e.g., number, order, etc.)
> model information
M
Information about the ML models to be used in each stage (e.g., identifiers, versions, etc.)
> usage information
O
Information about the planned usage of the split operation (e.g., inputs frequency/size, output frequency/size, etc.)
> number of nodes
O
Minimum number of nodes required to support AIML operation splitting
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.3 Split operation pipeline discovery response
| Table 8.14.3.3-1 shows the response sent by sent by the AIMLE server after processing of split operation pipeline discovery request.
Table 8.14.3.3-1: Split operation pipeline discovery response
Information element
Status
Description
Successful response
O
(NOTE 2)
Indicates that the request was successful.
> List of nodes (NOTE 1)
O
The list of discovered nodes.
> List of split operation profiles (NOTE 1)
O
The list of split operation profiles as specified in Table 8.14.3.3-2.
Failure response
O
(NOTE 2)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE 1: Only one of the IE shall be present for successful response.
NOTE 2: One of the IEs shall be present.
Table 8.14.3.3-2: Split operation profile
Information element
Status
Description
Split operation pipeline identifier
M
The identifier of the split operation pipeline.
Head endpoint
M
The endpoint information of the head node for providing intermediate data.
Tail endpoint
M
The endpoint information of the tail node for obtaining results.
Usage information
O
The usage information of the AI/ML split operation (e.g., inputs frequency/size, output frequency/size, etc.).
Stage information
M
The list of stage(s) of the AI/ML split operation, the nested information is provided for each stage.
> Stage identifier
M
The identifier of the stage.
> number of nodes
O
Number of nodes included in the stage
> head node
M
Endpoint of the head node - for providing initial the inference data.
> tail node
M
Endpoint of the head node - for obtaining inference results.
> order of the nodes
M
The order of the nodes in the AI/ML split operation (including head node, tail node).
>> list of nodes
M
List of all discovered node in the order in which they process the data.
>> notification target
O
Endpoint information where the inference result of the split operation is sent by the nodes
> Model information
M
The ML model information used in the stage (e.g., identifiers, versions, etc.).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.4 Split operation pipeline create request
| Table 8.14.3.4-1 shows the request sent by a AIMLE client to the AIMLE server for split operation pipeline create request.
Table 8.14.3.4-1: Split operation pipeline create request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL client ID, AIMLE client ID, UE identifier).
Security credentials
M
The security credentials of the requestor.
Split operation requirements
M
Split operation requirements.
> model information
M
Information about the ML models to be used in each stage (e.g., identifiers, versions, etc.).
> usage information
O
Information about the planned usage of the split operation (e.g., inputs frequency/size, output frequency/size, etc.).
> notification target
O
Endpoint information where the result of the split operation is sent by the tail node.
> stage information
M
Information about the split operation stages (e.g., number, order, etc.).
>> head node
M
Endpoint of the head node - for providing initial the inference data.
>> tail node
M
Endpoint of the head node - for obtaining inference results.
>> node information
M
List of Information about the nodes in order of the stage (e.g., identifier, endpoint, etc.).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.5 Split operation pipeline create response
| Table 8.14.3.5-1 shows the response sent by a AIMLE server after processing of split operation pipeline create request.
Table 8.14.3.5-1: Split operation pipeline create response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
> Split operation profile
O
The split operation profile as specified in Table 8.14.3.3-2.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.6 Split operation node register request
| Table 8.14.3.6-1 shows the request sent by a VAL server to the AIMLE server for split operation node register request.
Table 8.14.3.6-1: Split operation node register request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL server ID).
Security credentials
M
The security credentials of the requestor.
Node information
M
Information about the VAL server node (e.g., identifier, endpoints, etc.).
Split operation capabilities
M
Split operation capabilities of the VAL server.
> Model information
M
Information about ML model capabilities of the VAL server for split operation (e.g., identifiers, versions, etc.).
> Usage information
O
Information about usage capabilities of the VAL server for split operation (e.g., inputs frequency/size, output frequency/size, etc.).
Expiration time
O
The proposed expiration time of the registration.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.7 Split operation node register response
| Table 8.14.3.7-1 shows the response sent by an AIMLE server after processing of split operation node register request.
Table 8.14.3.7-1: Split operation node register response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
> Registration identifier
M
The identifier of the registration.
> Expiration time
O
The expiration time of the registration. To maintain an active registration, a registration update is required before the expiration time.
If the Expiration time IE is not included, it indicates that the registration never expires.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.8 Split operation node registration update request
| Table 8.14.3.8-1 shows the request sent by a VAL server to the AIMLE server for split operation node registration update request.
Table 8.14.3.8-1: Split operation node registration update request
Information element
Status
Description
Registration identifier
M
The identifier of the registration.
Security credentials
M
The security credentials of the requestor.
Node information
O
Information about the VAL server node (e.g., identifier, endpoints, etc.).
Split operation capabilities
O
Split operation capabilities of the VAL server.
> Model information
O
Information about ML model capabilities of the VAL server for split operation (e.g., identifiers, versions, etc.).
> Usage information
O
Information about usage capabilities of the VAL server for split operation (e.g., inputs frequency/size, output frequency/size, etc.).
Expiration time
O
The proposed expiration time of the registration.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.9 Split operation node registration update response
| Table 8.14.3.9-1 shows the response sent by an AIMLE server after processing of split operation node registration update request.
Table 8.14.3.9-1: Split operation node registration update response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
> Expiration time
O
The expiration time of the registration. To maintain an active registration, a registration update is required before the expiration time.
If the Expiration time IE is not included, it indicates that the registration never expires.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.10 Split operation node de-register request
| Table 8.14.3.10-1 shows the request sent by a VAL server to the AIMLE server for split operation de-register request.
Table 8.14.3.10-1: Split operation node de-register request
Information element
Status
Description
Registration identifier
M
The identifier of the registration.
Security credentials
M
The security credentials of the requestor.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.11 Split operation node de-register response
| Table 8.14.3.11-1 shows the response sent by an AIMLE server after processing of split operation de-register request.
Table 8.14.3.11-1: Split operation de-register response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.12 Split operation subscribe request
| Table 8.14.3.12-1 shows the request sent by an AIMLE client or a VAL server to the AIMLE server for split operation subscribe request.
Table 8.14.3.12-1: Split operation subscribe request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL server ID).
Security credentials
M
The security credentials of the requestor.
Split operation pipeline identifier
M
Identifier of the split operation pipeline for which the requestor is subscribing.
Notification endpoint
M
The notification endpoint (e.g. URL/URI/IP address) where the notifications should be sent to.
Event identifier
M
The event identifier for the subscription:
- Split operation pipeline information
- Split operation assistance information
Discovery filters
O
The set of characteristics to determine matching split operation profiles or nodes (as detailed in Table 8.14.3.2-1).
Applicable for "Split operation pipeline information".
Assistance information
O
The assistance information for the subscription:
- Aggregate the collected assistance information from NEF, NWDAF, and/or ADAES to generate assistance information, e.g. Time (time point(s) or time window(s)) to deliver the task or data for the split operation.
- Perform inference using ML model to generate assistance information, e.g. achievable QoS with current configuration for task or data delivery, or suggestion of QoS for task or data delivery.
Applicable for "Split operation assistance information".
Expiration time
O
The proposed expiration time of the subscription.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.13 Split operation subscription response
| Table 8.14.3.13-1 shows the response sent by an AIMLE server after processing of split operation subscription request.
Table 8.14.3.13-1: Split operation subscription response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
> Subscription identifier
M
The identifier of the subscription.
> Expiration time
O
The expiration time of the subscription. To maintain an active subscription, a subscription update is required before the expiration time.
If the Expiration time IE is not included, it indicates that the subscription never expires.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.14 Split operation notification
| Table 8.14.3.14-1 shows the request sent by an AIMLE server to notify a subscriber of detected split operation events.
Table 8.14.3.14-1: Split operation notification
Information element
Status
Description
Subscription identifier
M
The identifier of the subscription.
Event identifier
M
The identifier for the detected event, either:
- Split operation pipeline information
- Split operation assistance information
Availability information
O
The split operation availability information.
Applicable for "Split operation availability" event.
> List of split operation profiles
O
The list of newly available split operation profiles as specified in Table 8.14.3.3-2.
> List of nodes
O
The list of newly available nodes.
Split operation pipeline information
O
The split operation pipeline information.
Applicable for "Split operation pipeline information" event.
> Sub-event
O
Indicates the possible sub-event.
- Created
- Updated
- Deleted
> Split operation profile
O
The split operation profile that the VAL server participates to as specified in Table 8.14.3.3-2.
Assistance information
O
The split operation assistance information.
Applicable for "Split operation assistance information" event.
> Delivery time
O
Time (time point(s) or time window(s)) to deliver the task or data for the split operation.
> Achievable QoS
O
The achievable QoS with current configuration for task or data delivery.
> Suggestion of QoS
O
The suggestion of QoS for task or data delivery.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.15 Split operation subscribe update request
| Table 8.14.3.15-1 shows the request sent by an AIMLE client or a VAL server to the AIMLE server for split operation subscription update request.
Table 8.14.3.15-1: Split operation subscription update request
Information element
Status
Description
Subscription identifier
M
The identifier of the subscription.
Security credentials
M
The security credentials of the requestor.
Notification endpoint
O
The notification endpoint (e.g. URL/URI/IP address) where the notifications should be sent to.
Event identifier
O
The event identifier for the subscription:
- Split operation pipeline information
- Split operation assistance information
Split operation discovery filters
O
The set of characteristics to determine matching split operation profiles or nodes (as detailed in Table 8.14.3.2-1).
Applicable for "Split operation availability" event.
Expiration time
O
The proposed expiration time of the subscription.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.16 Split operation subscribe update response
| Table 8.14.3.16-1 shows the request sent by an AIMLE server after processing of split operation subscription update request.
Table 8.14.3.16-1: Split operation subscription update response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
> Expiration time
O
The expiration time of the subscription. To maintain an active subscription, a subscription update is required before the expiration time.
If the Expiration time IE is not included, it indicates that the subscription never expires.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.17 Split operation unsubscribe request
| Table 8.14.3.17-1 shows the request sent by an AIMLE client or a VAL server to the AIMLE server for split operation unsubscribe request.
Table 8.14.3.17-1: Split operation unsubscribe request
Information element
Status
Description
Subscription identifier
M
The identifier of the subscription.
Security credentials
M
The security credentials of the requestor.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.18 Split operation unsubscribe response
| Table 8.14.3.18-1 shows the response sent by an AIMLE server after processing of split operation unsubscribe request.
Table 8.14.3.18-1: Split operation unsubscribe response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
Failure response
O
(NOTE)
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of the IEs shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.19 Split operation pipeline update request
| Table 8.14.3.19-1 shows the request sent by an AIMLE client to the AIMLE server for split operation pipeline update request.
Table 8.14.3.19-1: Split operation pipeline update request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL client ID, AIMLE client ID, UE identifier).
Security credentials
M
The security credentials of the requestor.
Split operation pipeline identifier
M
The identifier of the AI/ML split operation pipeline.
Split operation pipeline information
M
Split operation pipeline information to be updated; this may include information elements defined in the split operation profile in Table 8.14.3.3-2.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.20 Split operation pipeline update response
| Table 8.14.3.20-1 shows the response sent by an AIMLE server after processing of split operation pipeline update request.
Table 8.14.3.20-1: Split operation pipeline update response
Information element
Status
Description
Successful response (NOTE)
O
Indicates that the request was successful.
> Split operation profile
O
The split operation profile as specified in Table 8.14.3.3-2.
Failure response (NOTE)
O
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of these IE is included.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.21 Split operation pipeline delete request
| Table 8.14.3.21-1 shows the request sent by an AIMLE client to the AIMLE server for split operation pipeline delete request.
Table 8.14.3.21-1: Split operation pipeline delete request
Information element
Status
Description
Requestor identifier
M
The identity of the requestor (e.g., VAL client ID, AIMLE client ID, UE identifier).
Security credentials
M
The security credentials of the requestor.
Split operation pipeline identifier
M
The identifier of the AI/ML split operation pipeline.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.14.3.22 Split operation pipeline delete response
| Table 8.14.3.22-1 shows the response sent by an AIMLE server after processing of split operation pipeline delete request.
Table 8.14.3.22-1: Split operation pipeline delete response
Information element
Status
Description
Successful response (NOTE)
O
Indicates that the request was successful.
Failure response (NOTE)
O
Indicates that the request failed.
> Failure cause
O
Indicates the failure cause.
NOTE: One of these IE is included.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15 AIMLE data management assistance
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.1 General
| AIMLE data management assistance is the process of the AIMLE server assisting AIMLE service consumers with managing data operations performed by VAL clients. The data operations include data preparation and data analysis. The AIMLE server offloads the AIMLE service consumer from interacting with AIMLE clients to manage the data operations. The VAL clients perform the actual data preparation and data analysis operations and send the outputs to the AIMLE server for aggregation.
The following clauses specify procedures, information flows, and APIs for AIMLE data management assistance.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.2 Procedure
| Pre-conditions:
1. AIMLE clients have registered with AIMLE server.
2. UE application data for AI/ML operations have been collected and dataset identifier have been assigned to each dataset. EVEX mechanism can be reused for data collection as described in 3GPP TS 26.531 [10].
Figure 8.15.2-1: AIMLE data management assistance
1. An AIMLE service consumer (e.g., VAL server) makes a request for AIMLE data management assistance. The request includes information as described in Table 8.15.3.1-1.
To prepare data for AIML operations, the AIMLE service consumer configures Data management operations to data preparation. Data preparation is performed to process data into a format that is required by ML models for the models to function properly, e.g.; before training or inferencing. The AIMLE service consumer includes data preparation requirements as described in Table 8.15.3.1-2.
To analyse data for AIML operations (i.e., to perform exploratory data analysis or EDA), the AIMLE service consumer configures Data management operations to data analysis. The AIMLE service consumer includes data analysis requirements as described in Table 8.15.3.1-3.
In addition to providing the data management requirements, the AIMLE service consumer provides either a list of AIMLE clients (if known) or AIMLE client selection criteria. When providing the AIMLE client selection criteria, the AIMLE service consumer indicates that the AIMLE server performs AIMLE client selection to select AIMLE clients for the data management operations.
2. The AIMLE server authenticates and authorizes the request. If authorized, the AMILE server assigns an identifier for the subscription.
3. The AIMLE server sends an AIMLE data management assistance subscription response to the AIMLE service consumer. The response includes information as described in Table 8.15.3.2-1.
4. The AIMLE server sends Client data processing trigger requests to AIMLE clients. The request can be for data preparation or data analysis. The request includes information as described in Table 8.15.3.4-1. If AIMLE client selection criteria were provided in step 1, the AIMLE server performs AIMLE client selection to select AIMLE clients that have Dataset ID and Dataset feature ID indicated in the data management requirements.
5. Each AIMLE client sends the requirements to trigger data management for the VAL client to perform the requested data operation. The VAL client performs the operation locally.
NOTE: The data preparation and data analysis functions operate in a similar manner as ML models. The AIMLE layer is able to transport the functions to the VAL clients and the VAL clients performs the associated function on the application data identified by the Dataset ID and Dataset feature ID as part of data preparation or data analysis in a similar manner as ML training.
6. After the data operation completes, each AIMLE client sends a response to the AIMLE server with information as described in Table 8.15.3.5-1.
7. The AIMLE server aggregates the output of each AIMLE client. Aggregation includes combining or performing a statistical operation on the received data. Data received from the AIMLE clients can be numerical or categorical, which allows the AIMLE server to aggregate the output data from the AIMLE clients. If necessary, steps 4–7 is repeated to complete all the required data operations.
8. The AIMLE server sends an AIMLE data management assistance notification to the AIMLE service consumer with information as described in Table 8.15.3.3-1.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3.1 AIMLE data management assistance subscription request
| Table 8.15.3.1-1 shows the request sent by an AIMLE service consumer to an AIMLE server for the AIMLE data management assistance subscription procedure.
Table 8.15.3.1-1: AIMLE data management assistance subscription request
Information element
Status
Description
Requestor identifier
M
The identifier of the requestor.
Data management operations
M
An indicator showing what data management type is being requested: data preparation, data analysis.
Data management requirements
M
Requirements for the data management request:
>Data preparation requirements
O
(NOTE 1)
Data preparation requirements as detailed in Table 8.15.3.1-2.
>Data analysis requirements
O
(NOTE 1)
Data analysis requirements as detailed in Table 8.15.3.1-3.
AIMLE clients list
O
(NOTE 2)
A list of AIMLE clients for which data management should be performed. The list may be specified by AIMLE client set identifier.
AIMLE client selection criteria
O
(NOTE 2)
Selection criteria for finding suitable AIMLE clients for AI/ML operations as detailed in Table 8.8.3.1-2.
NOTE 1: At least one of the information elements shall be provided.
NOTE 2: At least one of the information elements shall be provided.
Table 8.15.3.1-2: Data preparation requirements
Information element
Status
Description
Dataset identifier
M
An identifier for the dataset.
Data preparation requirements
M
Requirements for data preparation.
> Dataset ID
M
The identifier for the dataset.
> Dataset feature ID
M
Identifier or name of dataset feature to process.
> Data preparation function
M
This indicates the function which prepares the data, and it could be an identifier of a function if the function is available locally at the UE or an executable included in the request.
Table 8.15.3.1-3: Data analysis requirements
Information element
Status
Description
Dataset identifier
M
An identifier for the dataset.
Dataset analysis requirements
M
Requirements for data analysis.
> Dataset ID
M
The identifier for the dataset.
> Dataset feature ID
M
Identifier or name of dataset feature.
> Data analysis function
M
This indicates the function which performs the data analysis, and it could be an identifier of a function if the function is available locally at the UE or an executable included in the request.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3.2 AIMLE data management assistance subscription response
| Table 8.15.3.2-1 shows the response sent by the AIMLE server to the AIMLE service consumer for the AIMLE data management assistance subscription procedure.
Table 8.15.3.2-1: AIMLE data management assistance subscription response
Information element
Status
Description
Status
M
The status for the data management operation
Subscription identifier
M
An identifier for the subscription.
Expiration time
O
Expiration time for the subscription
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3.3 AIMLE data management assistance notify
| Table 8.15.3.3-1 shows the notification sent by the AIMLE server to the AIMLE service consumer for the AIMLE data management assistance subscription procedure.
Table 8.15.3.3-1: AIMLE data management assistance notify
Information element
Status
Description
Status
M
The status for the data management operation
Aggregated data preparation outputs
O
(NOTE 1)
(NOTE 2)
Provides outputs for data preparation: dataset identifier, dataset features, and prepared data output.
Aggregated data analysis outputs
O
(NOTE 1)
(NOTE 2)
Provides outputs for data analysis: dataset identifier, statistical outputs for each feature, list of outlier and anomaly values, and feature correlation information.
Timestamp
O
Timestamp of the data management notification
NOTE 1: At least one of the information elements shall be provided in the output.
NOTE 2: The output format can be numerical or categorical. If categorical, the format can be nominal or ordinal.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3.4 Client data processing trigger request
| Table 8.15.3.4-1 shows the request sent by the AIMLE server to AIMLE clients for the AIMLE client data processing trigger procedure.
Table 8.15.3.4-1: Client data processing trigger request
Information element
Status
Description
Requestor identifier
M
The identifier of the requestor.
Data management type
M
An indicator showing what data management type is being requested: data preparation, data analysis.
Data management requirements
M
Requirements for the data management request:
>Data preparation requirements
O
(NOTE)
Data collection requirements as detailed in Table 8.15.3.1-2.
>Data analysis requirements
O
(NOTE)
Data collection requirements as detailed in Table 8.15.3.1-3.
Operational schedule
O
A schedule to perform the requested data management operation.
NOTE: At least one of the information elements shall be provided.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.15.3.5 Client data processing trigger response
| Table 8.15.3.5-1 shows the response sent by AIMLE clients to the AIMLE server for the Client data processing trigger procedure.
Table 8.15.3.5-1: Client data processing trigger response
Information element
Status
Description
Status
M
The status for the data management operation
Data preparation outputs
O
(NOTE 1)
The output data after performing data preparation. One output data is generated for each requirement. (NOTE 2)
Data analysis outputs
O
(NOTE 1)
The output data generated by data analysis. One output data is generated for each requirement. (NOTE 2)
Timestamp
O
Timestamp of the data management operation
NOTE 1: At least one of the information elements shall be provided in the output.
NOTE 2: The output format can be numerical or categorical. If categorical, the format can be nominal or ordinal. The AIMLE server is able to process either type of data.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16 Support for Transfer Learning enablement
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.1 General
| This clause provides the procedures for the transfer learning (TL) enablement service, including the server-triggered procedure (in clause 8.16.2) and the client-triggered procedure (in clause 8.16.3).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.2 Procedure for server-triggered transfer learning enablement
| Figure 8.16.2-1 illustrates the procedure where the TL enablement is performed based on the request for either an ML task from VAL layer or for an analytics task from ADAES. Such TL enablement allows the consumer to discover the similar ML models to be used as base models for the TL, as well as to support the selection of the best model to be used as pre-trained model.
Pre-conditions:
1. VAL server is connected to AIMLE Server.
Figure 8.16.2-1: Procedure for Transfer Learning enablement
1. The VAL server sends a Transfer Learning model selection assistance request message to the AIMLE server to provide support for discovering and selecting the appropriate pre-trained model for a given ML task (for ADAE analytics ID or for a certain ML model ID).
2. The AIMLE Server discovers the possible entities which can provide a pre-trained model for this request. Such entities can be VAL servers or other ADAES or other AIMLE servers.
3. The AIMLE Server requests one or more pre-trained ML models which can be used for transfer learning for the target ML task. The ML repository identifies the base ML model as a pre-trained ML model that can be mapped to the target ML task and sends information on the models to AIMLE server which are candidate as pre-trained models available for the target task. This step reuses the ML model information discovery procedure as in clause 8.11.3.
4. The AIMLE Server evaluates with the support of the ML model repository, whether the pre-trained models are applicable to the ML task (ADAE analytics ID or model ID). This can be assisted using historical data or ML model rating based on previous utilization of these models for the certain ML task. Based on the evaluation (which can be based on the rating), the AIMLE Server determines one or more pre-trained models to be used for the ML task.
NOTE: In this step, the AIMLE Server can rate or set a weight to the pre-trained model or the source of the model.
5. The AIMLE Server sends to the VAL server a transfer learning selection assistance response to the VAL server, including the information for the pre-trained models (e.g., model ID, profile) which are identified for the ML task. Also, this may include the rating/weight for the pre-trained model if the VAL server needs to select among a list of them.
6. Based on the selected pre-trained model information, the VAL server retrieves the selected ML model using the procedure as in clause 8.2.2.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.3 Procedure for client-triggered transfer learning enablement
| Figure 8.16.3-1 illustrates the procedure where the TL enablement is performed for a VAL UE task which is a UE analytics task (e.g. ADAEC-provided analytics). Such TL enablement allows the VAL UE to perform ML model training using a pre-trained model from the server side and is beneficial for minimizing the computational load at the VAL UE side.
Pre-conditions:
1. AIMLE client is connected to AIMLE Server.
Figure 8.16.3-1 Procedure for client-triggered Transfer Learning enablement
1. The AIMLE client determines a requirement for using a pre-trained ML models for transfer learning for a VAL UE triggered ML task (e.g., UE analytics event).
2. The AIMLE client sends a UE transfer learning model selection assistance request to the AIMLE server to receive one or more pre-trained ML models which can be used for transfer learning for the target UE-triggered ML task.
3. The AIMLE Server requests one or more pre-trained ML models which can be used for transfer learning for the target ML task. The ML repository identifies the base ML model as a pre-trained ML model that can be mapped to the target ML task and sends information on the models to AIMLE server which are candidate as pre-trained models available for the target task.
4. The AIMLE Server sends a UE transfer learning model selection assistance response to the AIMLE client which includes information on the ML models which are candidate as pre-trained models available for the target UE ML task. Such models may be pre-trained for an ADAES analytics task (e.g., VAL server performance analytics) and are applicable to be used for the VAL UE analytics task (e.g. VAL session performance analytics).
5. The AIMLE client evaluates whether the pre-trained models are applicable to the target ML task. Based on the evaluation, the AIMLE client selects a base model to be used as pre-trained model for the ML task.
6. Based on the selected pre-trained model information, the AIMLE client retrieves the selected ML model using the procedure as in clause 8.2.2.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.4 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.4.1 Transfer learning model selection assistance request
| Table 8.16.4.1-1 shows the request sent by the VAL server to AIMLE server for the server-triggered transfer learning support procedure.
Table 8.16.4.1-1: Transfer learning model selection assistance request
Information element
Status
Description
Requestor identity
M
Identity of the VAL server performing the request.
VAL service identity
M
The identity of the VAL service for which the request applies.
ML task identity
O
(NOTE)
The ML task for which the transfer learning is to be used.
ADAE analytics ID
O
(NOTE)
The ADAES analytics ID (as specified in TS 23.436) for which the transfer learning is to be used, in case when transfer learning is used per analytics task.
ML model profile
O
(NOTE)
The ML model profile for which the transfer learning is to be used.
Transfer learning criteria
M
The criteria for identifying and selecting one or more pre-trained ML models. Such criteria include:
• the required feature(s) of a pre-trained model.
• training data requirements.
• type of transfer learning.
• the environment associated with the target ML task.
• permissions / restrictions for the pre-trained model.
ML model requirement information
O
Identifies the requirement for selecting a base model to be trained as pre-trained.
List of VAL UE IDs
O
List of VAL UEs associated with the ML model task
Model rating requirement
O
Identifies the requirement for providing rating of the ML model(s) to serve as pre-trained model.
NOTE: At least one of these IE shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.4.2 Transfer learning model selection assistance response
| Table 8.16.4.2-1 shows the response sent by the AIMLE server to the VAL server for the server-triggered transfer learning support procedure.
Table 8.16.4.2-1: Transfer learning model selection assistance response
Information element
Status
Description
Success response
O
(NOTE)
Indicates that the transfer learning model selection assistance request was successful.
> List of ML models
O
List of ML models selected by AIMLE server for training as candidate pre-trained model.
>> ML repository identifier and address
O
Provides the ID and address of the ML repository which stores the pre-trained ML model selected by AIMLE server for training as pre-trained model.
>> ML model information
O
Information on the selected model, specified in Table 8.11.4.1-2.
>> ML model rating
O
If requested, a rating parameter for the ML model to serve as pre-trained. Such rating can be based on the ML task similarity score e.g. based on the feature.
Failure response
O
(NOTE)
Indicates that the transfer learning model selection assistance request was failure.
> Cause
M
Reason for the failure.
NOTE: Only one of these information elements shall be present
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.4.3 UE transfer learning model selection assistance request
| Table 8.16.4.3-1 shows the request sent by the AIMLE client to AIMLE server for the client-triggered transfer learning support procedure.
Table 8.16.4.3-1: UE transfer learning model selection assistance request
Information element
Status
Description
VAL UE identity
M
Identity of the VAL UE (VAL client ID or AIMLE client ID) performing the request.
VAL service identity
M
The identity of the VAL service for which the request applies.
ML task identity
O
(NOTE)
The ML task for which the transfer learning is to be used.
ADAE analytics ID
O
(NOTE)
The ADAE analytics ID (as specified in 3GPP TS 23.436) for which the transfer learning is to be used, in case when transfer learning is used per UE analytics task.
ML model profile
O
(NOTE)
The ML model profile for which the transfer learning is to be used.
Transfer learning criteria
M
The criteria for identifying and selecting one or more pre-trained ML models. Such criteria include:
• the required feature(s) of a pre-trained model.
• training data requirements.
• type of transfer learning.
• the environment associated with the target ML task.
• permissions / restrictions for the pre-trained model.
ML model requirement information
O
Identifies the requirement for selecting a base model to be trained as pre-trained.
Model rating requirement
O
Identifies the requirement for providing rating of the ML model(s) to serve as pre-trained model.
NOTE: At least one of these IE shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.16.4.4 UE transfer learning model selection assistance response
| Table 8.16.4.4-1 shows the response sent by the AIMLE server to the AIMLE client for the client-triggered transfer learning support procedure.
Table 8.16.4.4-1: UE transfer learning model selection assistance response
Information element
Status
Description
Success response
O
(NOTE)
Indicates that the UE transfer learning model selection assistance request was successful.
> List of ML models
O
List of ML models selected by AIMLE server for training as pre-trained model.
>> ML model information
O
Information on the selected model, specified in Table 8.11.4.1-2.
>> ML model rating
O
If requested, a rating parameter for the ML model to serve as pre-trained.
Failure response
O
(NOTE)
Indicates that the UE transfer learning model selection assistance request was failure.
> Cause
M
Reason for the failure.
NOTE: Only one of these information elements shall be present
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17 Support for FL member grouping
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.1 General
| This clause provides the procedure for the grouping of the FL members using AIMLE. Such grouping for the given FL process, applicable to a specific VAL request or ML model ID or ADAE analytics ID. This grouping can be applicable also for a given service area in which one or more FL processes are expected to run. The grouping of FL members is performed for optimizing the process of selection and updating FL members which are entering or leaving the group, since due to dynamicity of FL member (e.g. AIMLE clients) changes this would be impose additional signalling / complexity.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.2 Procedure
| In this procedure, the AIMLE support capability is described for grouping the FL members, where the grouping is tailored to a specific ML task (VAL triggered task or analytics event/ID).
The grouping procedure covers the:
- creation of the FL member group;
- query for an individual FL Member whether it is part of the created group;
- change of the FL member group, including
a) modification of the group member based on a change on the availability or capability of the FL member (e.g. due to high load or energy consumption the FL member may have limited capability to act as FL client for a given area and time);
b) update of the group due to a new member entering or an existing member leaving the FL group; and
- deletion of the FL member group, and notification of the FL members regarding the deletion of the FL member group.
Figure 8.17.2-1 illustrates the procedure for supporting the FL member grouping.
Pre-conditions:
1. VAL Server is connected to AIMLE Server.
2. The candidate/selected FL member has registered to the FL member registry based on the capability in clause 8.4.
Figure 8.17.2-1: Procedure for FL member grouping
1. The VAL server sends an FL member grouping support request to the AIMLE Server for supporting an FL process. The initial request is to create the FL member grouping support as described in Table 8.17.3.1-1, which may be followed by other requests for querying (as described in Table 8.17.3.1-2), change (as described in Table 8.17.3.1-3), or deletion (as described in Table 8.17.3.1-4) of the FL member grouping support.
NOTE 1: The FL member grouping support request can be triggered by the AIMLE Server itself.
2. The AIMLE Server based on the request, determines the need for creating and processing a group consisting of the needed FL members for a given ML task (i.e., an ML model training/inference job ID). The need for creating an FL member group may be based on an ML task for a given AIMLE service area or for a given AIMLE service area where one or more ML tasks are expected to run, based on step 1 request.
3. To create, query, or change the FL member grouping support, the AIMLE Server fetches the available FL members for the given ML task (i.e., an ML model training/inference job ID) from the ML repository. Based on this information, AIMLE Server may select one or more FL members for the group for the ML task.
NOTE 2: If the FL member is an AIMLE client, this step re-uses procedure 8.9.2 on AIMLE client selection.
4. The AIMLE Server creates, configures, and processes the FL member group based on the available or selected FL members by the aggregator which may be the VAL Server or the AIMLE Server, generate group ID and store mapping to group member IDs. The criteria for determining the group members can be the capabilities of the FL participants, or whether the candidate participants are fixed or mobile nodes and their availabilities, the proximity of the participants among them.
For the creation of the FL member group, AIMLE may utilize SEAL GMS capability for the FL member group ID generation.
5. The AIMLE Server interacts with the each candidate FL member from the configured FL member group as shown in step 5a.
NOTE 3: If the request is about the deletion of the FL member group, the FL member grouping indication includes notification for the deletion of the FL member group.
5a. The AIMLE Server sends indication to the candidate FL member (the AIMLE client which is deployed on UE) about the group ID and the group member identities for the ML model ID / ADAE analytics ID (based on the request in step 1.
5b. The candidate FL member sends to the AIMLE Server a FL grouping indication acknowledge.
6. The AIMLE Server sends an FL member grouping support response to the VAL request indicating the group creation or providing the query response, change or deletion (based on step 1 request) and the group information.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.3.1 FL member grouping support request
| Table 8.17.3.1-1 shows the request sent by the VAL Server to AIMLE Server for the FL member grouping procedure to create the FL member grouping support.
Table 8.17.3.1-1: FL member grouping support create request
Information element
Status
Description
Requestor identity
M
Identity of the VAL Server performing the request.
VAL service identity
O (NOTE)
The identity of the VAL service for which the request applies.
ML model ID
O (NOTE)
The model ID for which the request applies
ADAE analytics ID
O (NOTE)
The ADAES analytics ID (as specified in TS 23.436) for which the FL grouping is to be used, in case when FL process is used for a given analytics task.
ML task information
O
Information related to the ML task / job for which the FL grouping is used. Such task can be an FL training task or FL inference task. This information may include an ML task ID which may be an FL process ID or correlation ID.
ML model profile
O
The ML model profile for which the FL grouping is to be used.
List of candidate/selected FL member IDs
O
The list of candidate or selected FL member identities (if known by the VAL Server) which are to be used in the grouping.
If the FL member is a VAL UE, this is equivalent to the VAL UE ID.
> FL member type
O
The type of FL members (FL client, FL Server)
> FL member status
O
The status (selected, candidate) of the FL member.
NOTE: At least one of these information elements shall be present
Table 8.17.3.1-2 shows the request sent by the VAL Server to AIMLE Server for the FL member grouping procedure to query the FL member within the group.
Table 8.17.3.1-2: FL member grouping support query request
Information element
Status
Description
FL member group identity
M
Identity of the FL member group which queried
FL member identity
O
Information on the queried FL member to be queried.
Table 8.17.3.1-3 shows the request sent by the VAL Server to AIMLE Server for the FL member grouping procedure to change the FL member grouping support.
Table 8.17.3.1-3: FL member grouping support change request
Information element
Status
Description
FL member group identity
M
Identity of the FL member grouping support
FL member group change
M
Information on the change type for the FL member group
>FL member update
O
(NOTE)
Identifies the FL Members that are to be updated
>>Cause
O
The cause for the FL member group update (e.g. FL member enter or leave the group)
>FL member group modify
O
(NOTE)
Identifies the FL Members that are to be modified
>>Cause
O
The cause for the FL member group modifies (e.g. change FL member availability, capability or FL member information).
NOTE: At least one of these shall be present
Table 8.17.3.1-4 shows the request sent by the VAL Server to AIMLE Server for the FL member grouping procedure to delete the FL member grouping support.
Table 8.17.3.1-4: FL member grouping support delete request
Information element
Status
Description
FL member group identity
M
Identity of the FL member grouping support
Cause
O
Cause for the deletion of the group
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.3.2 FL member grouping support response
| Table 8.17.3.2-1 shows the response sent by the AIMLE Server to the VAL Server for the FL member grouping procedure to create the FL member grouping support.
Table 8.17.3.2-1: FL member grouping support create response
Information element
Status
Description
Success response
O
(NOTE)
Indicates that the FL process support request was successful.
> FL group identifier(s)
O
Identifies the AIMLE-created FL group for the FL process and ML task (FL training or inference).
> List of FL member IDs / addresses
O
Provides the ID and address of the FL members which are part of the FL group.
>> FL member information
O
Information on the FL members such as availability, constraints, role/type.
Failure response
O
(NOTE)
Indicates that the FL process support request was failure.
> Cause
M
Reason for the failure.
NOTE: Only one of these information elements shall be present
Table 8.17.3.2-2 shows the response sent to the VAL Server by the AIMLE Server for the FL member grouping procedure to query the FL member within the group.
Table 8.17.3.2-2: FL member grouping support query response
Information element
Status
Description
Success response
O
(NOTE)
Indicates that the FL process support query request was successful.
> List of FL member IDs / addresses
O
Provides the ID and address of the FL members which are part of the FL group.
>> FL member information
O
Information on the FL members such as availability, constraints, role/type.
Failure response
O
(NOTE)
Indicates that the FL process query support request was failure.
> Cause
M
Reason for the failure.
NOTE: Only one of these information elements shall be present
Table 8.17.3.2-3 shows the response sent to the VAL Server by the AIMLE Server for the FL member grouping procedure to change the FL member grouping support.
Table 8.17.3.2-3: FL member grouping support change response
Information element
Status
Description
Success response
O
(NOTE)
Indicates that the FL process support request was successful.
> FL group identifier(s)
O
Identifies the AIMLE updated or modified FL group for the FL process and ML task (FL training or inference).
> List of FL member IDs / addresses
O
Provides the ID and address of the update or modified FL members which are part of the FL group.
>> FL member information
O
Information on the updated or modified FL members such as availability, constraints, role/type.
Failure response
O
(NOTE)
Indicates that the FL process support request was failure.
> Cause
M
Reason for the failure.
NOTE: Only one of these information elements shall be present
Table 8.17.3.3-4 shows the response sent to the VAL Server by the AIMLE Server for the FL member grouping procedure to delete the FL member grouping support.
Table 8.17.3.2-4: FL member grouping support delete response
Information element
Status
Description
Result
M
Positive or negative acknowledgement for the deletion of the FL member group.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.3.3 FL grouping indication
| Table 8.17.3.3-1 shows the notification sent by the AIMLE Server to the FL members (AIMLE clients, VAL Servers) for the FL member grouping procedure.
Table 8.17.3.3-1: FL grouping indication
Information element
Status
Description
Requestor identity
M
Identity of the AIMLE Server performing the request.
VAL service identity
O (NOTE 1)
The identity of the VAL service for which the grouping indication applies.
ML model ID
O (NOTE 1)
The model ID for which the indication applies
ADAE analytics ID
O (NOTE 1)
The ADAES analytics ID (as specified in 3GPP TS 23.436) for which the FL grouping is to be used, in case when FL process is used for a given analytics task.
FL group identifier(s)
M
Identifies the AIMLE-created, changed FL group for the FL process.
> List of FL member IDs / addresses
O
Provides the ID and address of the FL members which are part of the FL group.
>> FL member information
O
Information on the FL members such as availability, constraints, role/type.
FL group deletion information
O
(NOTE 2)
Indication that the FL group is going to be deleted based on VAL server request
> Cause
O
Cause for the expected deletion of the FL members group (e.g., due to AI/ML service termination or group UE mobility to different service area).
> Expiration time
O
Indicates the expiration time of the FL group deletion (in case the deletion of the FL group is expected in future time instance). If the Expiration time IE is not included, it indicates that the deletion of the group is instant.
NOTE 1: At least one of these information elements shall be present.
NOTE 2: This IE is mandatory if the indication is related to an FL group deletion.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.17.3.4 FL grouping indication acknowledge
| Table 8.17.3.4-1 describes the information flow FL grouping indication acknowledge from the FL members (AIMLE clients which are deployed on UEs) to the AIMLE server.
Table 8.17.3.4-1: FL grouping indication acknowledge
Information element
Status
Description
Success response
O
(NOTE)
Acknowledgement of FL grouping indication.
Failure response
O
(NOTE)
Indicates that the FL grouping indication was failure.
> Cause
O
Reason for the failure.
NOTE: Only one of these information elements shall be present
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.18 Support Vertical FL
| This clause describes procedure for supporting VFL among application layer multiple UEs.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.18.1 General
| The following clauses specify procedures, information flows and APIs to support VFL among Application Layer multiple UEs.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.18.2 Procedure for supporting VFL
| Pre-conditions:
1. VFL members registered their AIMLE client profile to the AIMLE Server. The VFL members may update their status or information in their AIMLE client profile to the AIMLE Server.
2. The datasets of each of the UEs (where the AIMLE Clients (as VFL members) are deployed on) belong to more than one different data domains.
3. The VAL server has successfully subscribed/registered with the AIMLE server for model training notifications.
Figure 8.18.2-1: Procedure for supporting VFL
1. An AIMLE server receives an ML model training request from a VAL server as described in clause 8.3.2. If ML model training request is received and the AIMLE server determines to use VFL training (VFL training between different data domains), the procedure continues to step 2.
2. [Optional] The AIMLE server retrieves the indicated ML model in step 1 using the ML model retrieval procedure as described in clause 8.1. If the retrieved model is already trained and meets the machine learning requirements requested by the consumer, the procedure continues to step 7.
3. If AIMLE client selection criteria are provided in step 1, then the AIMLE server continuously monitors and selects AIMLE clients for the VFL training as described step 4 of the clause 8.13.2. If a number of the required AIML clients is provided, the AIMLE server ensures the requirement is met when selecting AIMLE clients for VFL training. If a list of AIMLE clients is provided in step 1, the AIMLE server selects the provided AIMLE clients for the VFL training.
4. The AIMLE Server gets VFL members information (the VFL members information may be included in the request in step 1 (Table 8.3.3.1-1) or be obtained through step 3. The AIMLE Server interacts with the VFL members (AIMLE Clients which are deployed on UEs) for each data domain for ML model training capability evaluation as described in clause 8.19.2.
5. The AIMLE Server determines the VFL members for this VFL training process based on the information received in step 4 and parameters received in step 1 (Table 8.3.3.1-1). The AIMLE Server determines VFL members for each data domain. The criteria used by the AIMLE Server include:
- Available data and minimum number of data samples for the same sample among the VFL members.
- Feature alignment of the sample/datasets with data labels among the VFL members.
- Available time of the VFL members for support the VFL training operations.
- Capability and minimum number of the VFL members for the VFL training operations.
- AIML model information for the VFL members and for the AIMLE Server.
6. The AIMLE Server coordinates the selected VFL members for VFL training. During VFL training process, the VFL members send intermediate results to the AIMLE Server, and the AIMLE Server responds to the VFL members with the updated information (e.g. gradients). The information from the AIMLE server can be used to update the model parameters maintained at each VFL member for the different data domains.
6a. The AIMLE server may report to the VAL server with the training status, that includes intermediate training results, The VAL server may adjust its request on the ML model training. If the VAL Server is providing data labels to complete the training, the VAL Server sends updated training parameters for the AIMLE Server to distribute to the VFL members. The updated training parameters apply for models of VFL members associated with each data domain.
NOTE: How AIMLE Server coordinates the selected VFL members to perform VFL training is out of scope for this release.
7. The AIMLE server sends a ML model training notification to the VAL server as described in clause 8.3.2. If the training schedule is not complete (e.g., there are remaining training rounds), the AIMLE server configures the next set of training schedules and steps 3 to 6 are repeated for the next training round.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19 ML Model Training Capability Evaluation
| This clause describes procedure for supporting ML model training capability evaluation for FL (e.g., HFL, VFL).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19.1 General
| The following clauses specify procedures, information flows and APIs to support ML model training capability evaluation for FL (e.g., HFL, VFL). The ML model training capability result can be used by the AIMLE server to select FL members for FL training process (e.g. HFL, VFL).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19.2 Procedure for ML model training capability evaluation
| Pre-conditions:
1. AIMLE server determines to use FL (e.g., HFL, VFL) training.
2. AIMLE Server knows the information of the FL (e.g., HFL, VFL) members (e.g., AIMLE Client which is deployed to a UE).
Figure 8.19.2-1: Procedure for ML model training capability evaluation
1. The AIMLE Server sends ML model training capability evaluation request to the FL members (AIMLE Clients which are deployed on UEs). The request message includes information as described in Table 8.19.3.1-1.
2. The FL members evaluate their capability and availability to join the FL training process. The FL members (AIMLE Clients which are deployed on UEs) run the test task contained in the request in step 1 and determine if it can join the FL process. For VFL, as part of the test task, data alignment between the datasets of the different domains are determined. The VAL server may also provide data labels for the data alignment.
NOTE: The procedures for data collection from UE need to take user consent into account.
3. The FL members (AIMLE Clients which are deployed on UEs) send ML model training capability evaluation response to the AIMLE Server. The response message contains the information as described in Table 8.19.3.2-1.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19.3.1 ML model training capability evaluation request
| Table 8.19.3.1-1 shows the request sent by AIMLE server to FL (e.g., HFL, VFL) members (AIMLE Clients which are deployed on UEs) for ML model training capability evaluation.
Table 8.19.3.1-1: ML model training capability evaluation request
Information element
Status
Description
Requestor identifier
M
The identifier of the requestor.
Available time
O
Requirement on available time for supporting FL operations.
Test task
O
The task for test ML model training capability.
AI/ML model and model parameter(s)
O
Information about the AI/ML model and model parameters for use in FL training. In VFL, AI/ML models for different data domains are provided.
Requirement on dataset
O
Requirements on dataset for FL training.
>Common feature ID(s)
O
(NOTE 1)
Identifier(s) of the required features common to the dataset of the different data domains.
>Data domain feature ID lists
O
(NOTE 1)
List of features for each data domain(s) of the datasets at the UE.
>Data source
O
(NOTE 2)
Data source for the FL training.
NOTE 1: If Requirement on dataset is provided, at least one of these IEs shall be present when for VFL.
NOTE 2: If Requirement on dataset is provided, the IE shall be present when for HFL.
NOTE: The detail content of the Test task and AI/ML model is up to implementation.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.19.3.2 ML model training capability evaluation response
| Table 8.19.3.2-1 shows the response sent by FL members (AIMLE Clients which are deployed on UEs) to AIMLE server for the ML model training capability evaluation.
Table 8.19.3.2-1: ML model training capability evaluation response
Information element
Status
Description
Status
M
The status for the evaluation: success, fail.
• Success means join the FL training process.
• Fail means not join the FL training process.
Test result
O
The test result of the ML model training capability evaluation. The "test result" shall be provided when the "status" is "success".
Fail reason
O
The reason of the ML model training capability evaluation fail. The "fail reason" shall be provided when the "status" is "fail".
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20 AIML service operations control and management procedure
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.1 General
| The control and management of the AIML services is an essential requirement for the applications to manage the AIML services like model training, inference, discovery etc.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.2 AIML service operations control and management procedure
|
Figure 8.20.1-1: AIML service operations control and management procedure
1. The VAL server sends AIML service operations control and management request to the AIML Enablement server as per the Table 8.20.3.1-1.
2-3. The AIMLE server determines the required service operation mode to manage the AIML service operation lifecycle based on the AIML service operation mode, AIML service operation information. For example, the AIMLE server can perform client discovery/selection using the procedure defined in 3GPP TS 23.482 clause 9.3 and model training using the procedure defined in 3GPP TS 23.482 clause 8.3. Based on the AIML client identifier the AIMLE server sends the AIML Enablement client service operation request as per Table 8.20.3.3-1.
The AIML service operation mode includes start and stop operation. Start indicates the initiation of the AIML service and stop indicates termination of the AIML service.
The AIML Enablement client receiving the AIML service operation mode performs the service operation mode for the AIML service operation. The AIMLE client can configure and monitor the AIML service operation as per the AIML service operation mode configuration. Based on the AIML service operation mode status reporting configuration (periodic or event-triggered), the AIML client reports the service operation mode status to the AIML Enablement server. The AIML Enablement client sends a response indicating the success or failure of the AIML Enablement client service operation as per Table 8.20.3.4-1.
4. The AIML Enablement server provides the AIML service operations control and management response to the VAL server. The message includes AIML service operation ID and the reporting status of the AIML service operation.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.3.1 AIML service operations control and management request
| Table 8.20.3.1-1 shows the request sent by an AIML service consumer (e.g., VAL server) to an AIMLE server for the AIML service operations control and management request.
Table 8.20.3.1-1: AIML service operations control and management request
Information element
Status
Description
Requestor identity
M
The identifier of the requestor (e.g., VAL server).
VAL service identifier
O
An identifier for the VAL service associated with the requestor.
AIMLE client identifier(s)
O
(NOTE)
Indicates the identifier(s) of AIML enablement client(s).
AIMLE client set identifier
O
(NOTE)
An identifier for the AIMLE client set.
AIML service operation identifier
O
Indicates the AIML service operation identifier to identify the AIML service. (e.g., model training id, ml task id)
AIML service operation information
O
Indicates AIML service operation information. It includes AIML service model container, URI of the model to fetch the model from a repository, AIML service aggregator URI to send model updates, AIML service operation optimization assistance like maximum convergence time
AIML service operation mode
M
Indicates the required AIMLE service operation modes like start, stop. The start mode defines the initiation of the AIML service. The stop mode is defined to stop the AIML operation.
AIML service operation mode configuration
O
Indicates the configuration of the AIML service operation modes. It includes network utilization (like stop the AIML service when latency is worse than x milliseconds, time limit threshold (like stop the AIML service after 24 hours), model performance (like stop the AIML service when model accuracy is 99% achieved)
AIML service operation mode status reporting
O
Indicates the reporting configuration of the AIML service operation status like periodic (e.g. time interval) or event based (e.g., transition of AIML service operation from stop to start)
NOTE: One of the information elements is present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.3.2 AIML service operations control and management response
| Table 8.20.3.2-1 shows the request sent by an AIML Enablement server to an AIML service consumer (e.g., VAL server) for the AIML service operations control and management response.
Table 8.20.3.2-1: AIML service lifecycle management response
Information element
Status
Description
VAL service identifier
O
An identifier for the VAL service associated with the requestor.
AIML service operation ID
M
An identifier to identify the AIML service operation ID
AIML service operation mode report status
M
Indicates the current state of AIMLE service operation. E.g., start, stop
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.3.3 AIML Enablement client service operation request
| Table 8.20.3.3-1 shows the request sent by an AIML Enablement server to an AIML Enablement client for the AIML Enablement client service operation request.
Table 8.20.3.3-1: AIML Enablement client service operation request
Information element
Status
Description
Requestor identity
M
The identifier of the requestor (e.g. AIML service consumer).
VAL service identifier
O
An identifier for the VAL service associated with the requestor.
AIML service operation ID
M
An identifier to identify the AIML service operation ID
AIML service operation mode
M
Indicates the required AIMLE service operation modes like start, stop.
AIML service operation information
O
Indicates AIML service operation information. It includes AIML service model container, URI of the model to fetch the model from a repository, AIML service aggregator URI to send model updates, AIML service operation optimization assistance like maximum convergence time
AIML service operation mode configuration
O
Indicates the configuration of the AIML service operation modes. It includes network utilization (like stop the AIML service when latency is worse than x milliseconds, time limit threshold (like stop the AIML service after 24 hours), model performance (like stop the AIML service when model accuracy is 99% achieved)
AIML service operation mode status reporting
O
Indicates the reporting configuration of the AIML service operation status like periodic (e.g. time interval) or event based (e.g. transition of AIML service operation from stop to start)
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.20.3.4 AIML Enablement client service operation response
| Table 8.20.3.4-1 shows the request sent by an AIML Enablement client to an AIML Enablement server for the AIML Enablement client service operation response or update response.
Table 8.20.3.4-1: AIML Enablement client service operation response
Information element
Status
Description
VAL service identifier
O
An identifier for the VAL service associated with the requestor.
AIML service operation ID
M
An identifier to identify the AIML service operation ID
AIML service operation mode status
M
Indicates the current state of AIMLE service operation. Possible values start, stop
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21 ML model update
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21.1 General
| This clause provides the procedures to support ML model re-training and update when model performance degradation is observed by the AIML enablement layer. The model update procedure also supports using an existing model to re-train the model using Transfer Learning. Additionally, if the degraded model is related to other models due to e.g., Transfer Learning, the AIMLE server may trigger the update of those related models as well.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21.2 Procedure
| Figure 8.21.2-1 depicts the procedure where the AIML enablement capability can trigger model update upon detecting model performance degradation.
Pre-conditions:
1. The AIMLE Server has provided a ML model to the AIMLE Consumer. The AIMLE consumer can be a VAL server, AIMLE client, or ADAE server.
2. The AIMLE Consumer detects a performance degradation of the ML model. If the consumer is an ADAE server, performance degradation may be detected as described in clause 8.17 in 3GPP TS 23.436 [4].
Figure 8.21.2-1: Support for ML model update
1. The AIMLE Consumer sends an ML model update request to the AIMLE Server that includes the ID of the model and performance degradation information.
2. Based on the performance degradation information, the AIMLE Server determines whether to update the ML model. If the AIMLE Server does not update the model, steps 3 and 4 are skipped.
3. The AIMLE Server retrieves the ML model information from the ML Repository as described in clause 8.11.3. The AIMLE Server may also perform ML model discovery to determine whether an existing ML model stored by the ML Repository can be used to replace the degraded model or train the new model (e.g., using Transfer Learning). If an existing model can be used to replace the degraded model, step 4 is skipped, and the identified model is provided in step 5.
The AIMLE Server can also discover models that are related due to Transfer Learning or the use of the same training data, to identify additional models that may require an update.
4. The AIMLE Server performs ML model re-training, which corresponds to the ML model training procedure as described in clause 8.3. The updated model is stored in the ML repository once re-training is complete.
If the degraded model is linked to other models (e.g., due to Transfer Learning, or the same training data has been used), the AIMLE Server may trigger the re-training and update of those related models.
5. The AIMLE Server provides the updated ML model to the AIMLE Consumer either by sending it directly, or by providing endpoint information to retrieve it from the ML Repository.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21.3.1 ML model update request
| Table 8.21.3.1-1 details the ML model update request IEs.
Table 8.21.3.1-1: ML model update request
Information element
Status
Description
Requestor Identity
M
The identity of the AIMLE Consumer sending the request.
ML model ID
M
Provides the ID of ML model for which the performance degradation has been detected.
Performance degradation information
O
Provides details about the detected performance degradation, such as the time, instances, or information on the degraded metrics (e.g. accuracy, recall, F1score).
ML model retrieval endpoint
O
The endpoint (e.g., URL, URI, IP address) where the ML model file can be retrieved.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.21.3.1 ML model update response
| Table 8.21.3.2-1 details the ML model update response IEs.
Table 8.21.3.2-1: ML model update response
Information element
Status
Description
Successful response
O
(NOTE 1)
Indicates that the model has been updated.
> ML model
O
(NOTE 2)
Provides the updated ML model.
> ML model retrieval endpoint
O
(NOTE 2)
The endpoint (e.g., URL, URI, IP address) where the ML model file can be retrieved.
> ML model information
O
Provides information of the ML model, specified in Table 8.11.4.1-2.
Failure response
O
(NOTE 1)
Indicates that the request has failed.
> Cause
O
Indicates the failure cause.
NOTE 1: Only one of these information elements shall be provided.
NOTE 2: At least one of these information elements shall be provided.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22 ML model performance monitoring
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.1 General
| The following clauses specify procedures, information flows, and APIs for ML model performance monitoring and potential degradation detection.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.2 Procedure
| Pre-conditions:
1. One or more AIMLE services (at the AIMLE server or clients) using the given ML model are ongoing.
Figure 8.22.2-1: ML model performance monitoring
1. The VAL server sends an ML model performance monitoring subscription request to AIMLE server, requesting to assist in monitoring the ML model performance. This request consists of ML model information and optionally the AIML service information, etc.
2. The AIMLE server checks whether the VAL server is authorized to perform the ML model performance monitoring request.
3. If the VAL server is authorized, AIMLE server returns the success response, otherwise a failure response indication the reason for failure.
4. The AIMLE server identifies the AIMLE services which are utilizing the requested ML model. Such AIMLE services can be an ML model training service or an HFL service at the AIMLE server or clients.
The identification of the AIMLE services may be performed via fetching ML model information from the ML repository using ML model management procedure as in clause 8.11.3.
Then, the AIMLE server then starts monitoring the AIMLE service performance (e.g. accuracy, a KPI or QoS metric related to the AIML operation). This step includes receiving information from one or more AIMLE clients performing an operation based on the target ML model (e.g., based on step 7 of procedure in clause 8.12.2 or step 6 of procedure in clause 8.15.2), with an expected or experienced deviation of the required performance of the AIMLE service.
5. The AIMLE server detects an expected ML model degradation (e.g., model drift, data drift) based on the deviation of the performance of the AIMLE service as indicated in step 4.
6. The AIMLE server based on the expected ML model degradation, it may also indicate and execute a trigger action to ensure meeting the AIMLE service requirement.
Such trigger action may be either an adaptation of the AIMLE service, such as training of a new ML model for the AIMLE by the same or a different AIMLE client, or re-training of the ML model by the same or different AIMLE client; or termination of the AIMLE service and initiating a new AIMLE service with a new ML model.
NOTE: If this action involves re-selecting an AIMLE client for the AIMLE service, this is based on the procedure defined in clause 8.9.
7. The AIMLE server notifies the VAL server on the expected ML model degradation and if requested the triggered adaptation of the AIMLE service.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.3.1 ML model performance monitoring subscription request
| Table 8.22.3.1-1 shows the request sent by a VAL server to an AIMLE server for the ML model performance monitoring.
Table 8.22.3.1-1: ML model performance monitoring subscription request
Information element
Status
Description
Requestor identity
M
The identifier of the requestor (e.g., VAL server).
ML model identifier
M
The identifier of the ML model for which the monitoring applies.
Notification endpoint
M
The notification endpoint (e.g. URL/URI/IP address) where the notifications should be sent to.
AIML operation information
O
The AIMLE operation (ML model training, HFL, VFL, TL) for which the ML model is used.
> VAL service ID
O
The VAL service identifier of the AIMLE service using the ML model (if known by the requestor).
> AIMLE client ID(s)
O
The identifier(s) of the AIMLE client(s) training the ML model (if known by the requestor).
> AIMLE service KPI
O
One or more KPIs for the AIMLE service performance (latency, accuracy, etc).
Monitoring report configuration
M
The reporting configuration for the monitoring service (thresholds for triggering a monitoring event, e.g. minimum accuracy, delay, whether the reporting is one time or periodical or event-triggered).
Area of interest
O
The geographical or service area for which the monitoring applies.
Time validity
O
The time validity for the monitoring subscription.
Trigger Action requirement
O
This requirement identifies policies for triggering an action based on a monitoring event (e.g. if degradation is detected, to train a new model or re-selecting AIMLE clients).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.3.2 ML model performance monitoring subscription response
| Table 8.22.3.2-1 shows the response sent by the AIMLE server to the VAL server for the ML model performance monitoring subscription.
Table 8.22.3.2-1: ML model performance monitoring subscription response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the ML model performance monitoring request was successful.
> Subscription ID
M
Subscription identifier corresponding to the subscription.
> Expiration time
O
Indicates the expiration time of the subscription. To maintain an active subscription, a subscription update is required before the expiration time.
Failure response
O
(NOTE)
Indicates that the request has failed.
> Cause
O
The cause for the request failure.
NOTE: Only one of these information elements shall be present.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.22.3.3 ML model performance monitoring notify
| Table 8.22.3.2-1 shows the notification sent by the AIMLE server to the VAL server for the ML model performance degradation.
Table 8.22.3.2-1: ML model performance monitoring notify
Information element
Status
Description
Subscription ID
M
Subscription identifier corresponding to the subscription.
ML model ID
M
Identity of the ML model.
ML model degradation indication
M
Identifies the degradation of the ML model.
>ML model degradation parameters
O
The performance metrics which are expected to be degraded (F1-score, recall, precision, accuracy).
> Cause
O
The cause for the degradation of the ML model.
Trigger Action
O
The trigger action, which is notified, and may be one of the following:
• the adaptation of the AIMLE service, such as training of a new ML model for the AIMLE by the same or a different AIMLE client,
• the re-training of the ML model by the same or different AIMLE client,
• the termination of the AIMLE service and initiating a new AIMLE service with a new ML model.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23 AIMLE assisted ML model selection
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.1 General
| ML model selection is an important consideration for successful ML training and deployment. Many ML models exist for ML applications and some models can generate better results than other models for a particular dataset. The procedure allows AIMLE service consumers to request assistance from an AIMLE server with the selection of appropriate ML models for a given dataset and for the provided requirements. The AIMLE server coordinates the selection of candidate ML models and training the ML models with the given dataset. A list of ML models with corresponding performance is return to the AIMLE service consumer.
The following clauses specify procedures, information flows, and APIs for ML model selection.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.2 Procedure
| Pre-conditions:
1. The AIMLE service consumer has identified datasets and ML requirements.
Figure 8.23.2-1: AIMLE assisted ML model selection
1. An AIMLE service consumer (e.g. VAL server) sends a subscription request to an AIMLE server. The request includes information as described in Table 8.23.3.1-1. The AIMLE service consumer provides a list of candidate ML models, dataset identifiers, and requirements for training the candidate ML models. The AIMLE service consumer can either provide AIMLE client set identifiers or AIMLE client selection criteria for selecting the AIMLE clients to train the candidate ML models.
2. The AIMLE server authenticates the requestor and checks authorization for the request. If authorized, the AMILE server assigns an identifier for the subscription.
3. The AIMLE server sends a ML model selection subscription response that includes information in Table 8.23.3.2-1.
4. The AIMLE server determines if additional ML models can be candidates for the type of ML application based on the ML model requirements provided in step 1 and selects additional candidate ML models to train with the given dataset. The AIMLE server can discover models from ML repository to determine the list of candidate ML models as described in clause 8.11.3.
5. The AIMLE server performs ML model training for each candidate model. ML model training can be for split AI/ML operation as described in clause 8.14, Transfer Learning as described in clause 8.16, or Federated Learning as described in clauses 8.12 and 8.18. If AIMLE client selection criteria were provided in step 1, the AIMLE server performs AIMLE client selection as described in clauses 8.9 or 8.13 during the training of the candidate ML models.
6. The AIMLE server performs ML model information storage as described in clause 8.11 for each trained ML model.
7. The AIMLE server aggregates and determines the performance of each ML model with the given dataset.
8. The AIMLE server sends a notification to the AIMLE service consumer and include information as described in Table 8.23.3.3-1. The notification includes a list of trained candidate ML models with corresponding model information and performance. The AIMLE service consumer can then select the best performing ML models from the list provided in the notification.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.3.1 AIMLE assisted ML model selection subscription request
| Table 8.23.3.1-1 shows the request sent by an AIMLE service consumer to an AIMLE server for the AIMLE assisted ML model selection subscription procedure.
Table 8.23.3.1-1: ML model selection subscription request
Information element
Status
Description
Requestor identifier
M
The identifier of the requestor.
AIML profile
M
Requirements for the ML model selection operation.
> Candidate ML models
M
A list of ML model identifiers (and initial model parameters) to train. The list provides candidate ML models to evaluate against the provided dataset.
> ML model requirements
O
ML model requirements for the AIMLE server to use for selecting additional candidate ML models for training with the provided datasets. The requirements can be any of the ML model information as described in Table 8.11.4.1-2.
> AIMLE client set identifiers
O
(NOTE1)
A list of AIMLE client set identifiers to train the ML model.
> AIMLE client selection criteria
O
(NOTE1)
Selection criteria for finding suitable AIMLE clients for training the ML model.
> Number of required AIMLE clients
O
(NOTE2)
A minimal number of AIMLE clients required for training the ML model.
> Dataset identifiers
M
Dataset identifiers to use for training and evaluating model performance to obtain a list of ML model rankings.
> Training requirements
M
Training requirements as detailed in Table 8.23.3.1-2.
Notification target
O
Endpoint information for receiving notifications.
Notification settings
O
Notification settings for which the AIMLE server provides ML model status: after, after certain job percentage completion, periodically based on date and time, upon error events, etc.
NOTE1: At least one of the information elements shall be provided.
NOTE2: Mandatory if AIMLE client selection criteria are present.
Table 8.23.3.1-2: Training requirements
Information element
Status
Description
Performance metric
M
Identifies the performance metric to evaluate ML model training. Performance metric can be mean absolute error, mean squared error, accuracy, precision and recall, etc. The performance metric indicates the performance of the ML model.
Performance target
O
A target performance that indicates acceptable performance has been reached and training can be stopped.
Number of training rounds
M
A minimum number of training rounds for the ML training.
Number of data samples
M
A minimum number of data samples for the ML training.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.3.2 ML model selection subscription response
| Table 8.23.3.2-1 shows the response sent by the AIMLE server to the AIMLE service consumer for the AIMLE assisted ML model selection subscription procedure.
Table 8.23.3.2-1: ML model selection subscription response
Information element
Status
Description
Status
M
The status for the ML model selection operation
Subscription identifier
M
An identifier for the subscription.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.23.3.3 ML model selection notification
| Table 8.23.3.3-1 shows the notification sent by the AIMLE server to the AIMLE service consumer for the AIMLE assisted ML model selection subscription procedure.
Table 8.23.3.3-1: ML model selection notification
Information element
Status
Description
Subscription identifier
M
The identifier for the subscription that notification is associated with.
Operational status
M
The status for the ML model selection operation. The status can represent the estimate percentage completion or associated with the notification settings.
Trained ML models
M
The results of the ML model training.
> ML model information
M
Information about the ML model such as the ML model type as described in Table 8.11.4.1-2.
> Model performance
M
The performance metric for training the ML model.
Elapse time
O
The time that has elapsed for the ML model selection operation.
Timestamp
O
Timestamp of the notification.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24 AIMLE context transfer
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24.1 General
| This clause describes AIMLE context transfer procedure between AIMLE servers (over AIML-E reference point).
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24.2 Procedure
| Pre-conditions:
1. Each edge AIMLE server manages the AIMLE clients within its service area to perform AI/ML operations.
2. A UE associated with an AIMLE client moves from a source service area (managed by a source edge AIMLE server) to a target service area (managed by a target edge AIMLE server). The transition triggers application context relocation (ACR) procedure between the two edge AIMLE servers (as source EAS and target EAS, respectively) as specified in 3GPP TS 23.558 [12].
Figure 8.24.2-1. AIMLE context transfer
1. The source edge AIMLE server sends an AIMLE context transfer request to the target AIMLE server as described in Table 8.24.3.1-1. The request includes AIMLE context information which is generated based on the responses/notifications received from the transitioned AIMLE client (e.g. step 7 of clause 8.12.2 or step 3 of clause 8.20.1).
The AIMLE context information is used by the target edge AIMLE server to determine whether the information (e.g. AI/ML operation output/result) received from the AIMLE client should be transferred to the source edge AIMLE server (or another edge AIMLE server that has been associated with the AIMLE client). For example, the target AIMLE server can forward the AIMLE service results received from the AIMLE client to the source AIMLE server if the results are only applicable to the source service area or the AIMLE client is part of a split operation pipeline formed in the source service area.
2. The target edge AIMLE server sends an AIMLE context transfer response to the source edge AIMLE server as described in Table 8.24.3.2-1.
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24.3 Information flows
| |
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24.3.1 AIMLE context transfer request
| Table 8.24.3.1-1 shows the AIMLE context transfer request that is sent by a source AIMLE server to a target AIMLE server.
Table 8.24.3.1-1: AIMLE context transfer request
Information element
Status
Description
Requestor identifier
M
The identifier of the requestor (e.g., AIMLE server).
AIMLE Context Information
M
The AIMLE context information as described in Table 8.24.3.1-2.
Table 8.24.3.1-2: AIMLE context information
Information element
Status
Description
AIMLE client ID
M
The identifier of the AIMLE client associated with the context.
Current managing AIMLE server
O
The identifier of the AIMLE server that is currently managing the AIMLE client, i.e. the AIMLE server associated with the service area that the AIMLE client is currently in.
Previous managing AIMLE server
O
List of identifiers of AIMLE servers that have been associated with the AIMLE client. The list is populated by adding the identifier of the source edge AIMLE server whenever the UE transitioned from a source edge area to a target edge area.
AIMLE service status
O
Status of the AIML operations (task) at the AIMLE client, e.g. “active”, “paused”, “completed”, percentage of completion.
AIMLE service results
O
Results of the AIML operations (task) performed by the AIMLE client.
AIMLE service applicability
O
Applicability information of the AIML operations performed by the AIMLE client, e.g. the operation results are applicable within a certain edge service area, the operations are applicable within a certain split operation pipeline.
>ML context information
O
Context information related to the ML operation that the AIMLE client is participating in or performing.
>> VAL service Information
O
Information related to the VAL service for which the AIMLE task is performed (e.g., the VAL service identifier for the AIMLE HFL training operation).
>> ML task
O
Type of ML task (model training, model testing, model inference, model transfer, model offload, model split, intermediate AI/ML operation/task) to be continued at the target AIMLE server.
>> ML task information
O
Information related to the ML task mentioned in “ML task” information element.
The Model Training task Information may include training objective to be achieved, HFL training information, VFL training information, data set information for training, status of training operation at AIMLE client (e.g. “active”, “paused”, “completed”), training results, etc.
The Model Inference task information may include Inference results, inference job id, etc.
The Model split task information may include split operation profile as specified in table 8.14.3.3-2.
>> ML model information
M
Model information related to the ML task. This information may include, the model identifier, Information to fetch ML model information, address (e.g., a URL or an FQDN) of the ML model file or address of the model repository where the ML model resides, Model parameters from ML training, etc)
|
29cb23e409b7480fb8bd6cba8e348433 | 23.482 | 8.24.3.2 AIMLE context transfer response
| Table 8.24.3.2-1 shows the AIMLE context transfer response that is sent by the target edge AIMLE server to the source edge AIMLE server.
Table 8.24.3.2-1: AIMLE context transfer response
Information element
Status
Description
Successful response
O
(NOTE)
Indicates that the request was successful.
Failure response
O
(NOTE)
Indicates that the request failed.
> Cause
O
Indicates the cause of request failure
NOTE: One of the IEs shall be present.
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.