Tech-invite3GPPspaceIETF RFCsSIP
Quick21222324252627282931323334353637384‑5x

Content for  TS 23.288  Word version:  17.5.0

Top   Top   Up   Prev   Next
1…   4…   5…   5A…   6…   6.1.4…   6.1.4.4…   6.1A…   6.1B…   6.1C   6.2…   6.2.3…   6.2.6…   6.2.6.2   6.2.6.3…   6.2.6.3.3   6.2.6.3.4   6.2.6.3.5   6.2.6.3.6   6.2.7…   6.2A…   6.2B…   6.3…   6.4…   6.5…   6.6…   6.7…   6.7.3…   6.7.4…   6.7.5…   6.8…   6.9…   6.10…   6.11…   6.12…   6.13…   6.14…   7…   8…   9…   10…

 

6.2.6  Enhanced Procedures for Data Collection |R17|p. 67

6.2.6.0  Generalp. 67

Data collection may be performed via DCCF, MFAF and ADRF, when such NFs are deployed, or via NWDAF (hosting DCCF and/or ADRF). The NF services used for DCCF, MFAF, ADRF and the NWDAF (hosting DCCF and/or ADRF) are listed in Table 6.2.6.0-1.
Service producer Service Reference
NWDAFNnwdaf_DataManagementclause 7.4
DCCFNdccf_DataManagement
Ndccf_ContextManagement
clause 8.2
clause 8.3
MFAFNmfaf_3daDataManagement
Nmfaf_3caDataManagement
clause 9.2
clause 9.3
ADRFNadrf_DataManagementclause 10.2
DCCF, MFAF, and NWDAF hosting DCCF shall use the same services listed in clause 6.2.1 from OAM, and NFs (including AFs directly or via NEF) to collect data. Additionally, the new services for data exposure from DCCF, MFAF, ADRF and NWDAF (hosting DCCF and/or ADRF) as specified in clause 8, clause 9, clause 10 and clause 7.4 may also be used for data collection.
The NWDAF, DCCF, ADRF shall obtain the proper information to perform data collection for a UE, a group of UEs or any UE following the principles of clause 6.2.1.
When NWDAF or DCCF need to discover the sources of data collection, they follow the principles defined in clause 6.2.2.1 in the case of data collection from NFs; in clause 5A.2 in the case of DCCF deployed in the network and requiring data from NWDAFs or ARDFs that independently collect data; clause 5.2 in the case of NWDAF (hosting DCCFs) and collecting data from other NWDAFs.
Up

6.2.6.1  Bulked Data Collectionp. 67

6.2.6.1.0  Generalp. 67
NWDAFs or DCCF or MFAF or ADRF may provide bulked data to consumers as an alternative to providing individual events (i.e. subscription to multiple event IDs to obtain the data required for an analytics generation).
The bulked data is the set of data samples from the collected event notifications to be used for an Analytics ID for a consumer of NWDAF, DCCF, MFAF or ADRF. A data sample may be a notification, in which case the bulked data may comprise a group of received notifications, or a data sample may be information extracted from a notification and processed, in which case the bulked data comprises the processed information. The bulked data can be used for the purpose of analytics inference or ML model training.
The bulked data is generated based on the set of data samples from event notifications collected from NFs/OAM and the properties for the selection of such data, Consumers of bulked data or operators may define different rules (e.g. aggregation formats or processes) for generation of bulked data for training or inference.
NFs capable to expose bulked data have the following capabilities:
  • Exposing runtime collected data (e.g. data from NFs/AFs/OAM retrieved via notification mechanisms), or historical collected data (e.g. data from NFs/AFs/OAM that were at some point collected, then stored), or both;
  • Applying selection processes of data samples or processing mechanisms for the generation of the bulked data according to bulked data formatting and processing instructions provided by consumers of the bulked data or defined by an operator. The bulked data formatting and processing instructions may include the formatting and processing instructions as specified in clause 5A.4 and further instructions as described in clause 6.2.6.1.1. Such instructions define the allowed and/or restricted properties and/or processes to be applied to the set of data from the collected event notifications to be used for the bulked data, the properties and processes being:
    • The properties associated with the Bulked data formatting and processing are filters over the data to be associated with the bulked data. The properties for bulked data are defined as per Data Specification parameters in clause 6.2.6.1.1.
    • The processes associated with the bulked data formatting and processing are mechanisms applied to the data to be associated with the bulked data and are defined according to the Formatting and Processing parameter defined in clause 5A.4 and the further instructions defined in clause 6.2.6.1.1. These processes may comprise: definition if data to be used for composing the bulked data is directly extracted from collected events, or the data is extracted from event notification of the same event type and pre-processed, or both; applying anonymization of data fields in the bulked data to avoid exposing undesired information, aggregation levels (i.e. per cell, per UEs, or temporal, e.g. per hours or days).
  • Having the mapping of the Service Operation that have to be used for collecting data of the bulked data associated with an Analytics ID.
Up
6.2.6.1.1  Services for Bulked Data Collectionp. 68
NWDAF, DCCF, MFAF, ADRF may expose service(s), respectively, the Nnwdaf_DataManagement_Subscribe or Ndccf_DataManagement_Subscribe, Nmfaf_3daDataManagement_Configure, Nadrf_DataManagement_RetrievalSubscribe service operation with a request for bulked data including the following input parameters:
  • Data Specification:
    • Event ID(s) or Analytics ID(s);
      • In the case of Event IDs, the Data Specification fields includes the fields Target of Event Reporting and Event Filter Information as defined in TS 23.502, clause 4.15.1, and Bulked Data Type parameter, which can be set to "raw data samples" (i.e. data is directly extracted from collected events) or "pre-processed data samples" (i.e. data from collected events is processed and the processed data is included in the bulked data) or a combination of both;
      • If the Analytics ID(s), the Data Specification fields contain:
        • Target of Reporting including a tuple with Analytics ID; Bulked Data Type, which can be set to "raw data samples" (i.e. data is directly extracted from collected events) or "pre-processed data samples" (i.e. data from collected events is processed and the processed data is included in the bulked data) or a combination of both; analytics stage (inference or training);
        • Filter Information may include fields related to the Analytics ID such as: Target of Analytics Information (e.g. any UE, list of UEs, groups of UEs); Analytics Filter Information (e.g. area of interest, DNN, Application, S-NSSAI). The Analytics ID also determines the Service Operation from NFs, OAM to be used and type of data (i.e. Event IDs, OAM measurements) to be collected and associated with the bulked data.
  • Service Operation in the case of Event ID, defines the service operation to be used by NWDAF, DCCF, MFAF, or ADRF to request data (e.g. Namf_EventExposure_Subscribe or OAM Subscribe)
  • Bulked Data Formatting and Processing: the parameters defined in TS 23.502, clause 4.15.1 for Event Reporting Information and Formatting and Processing as defined in clause 5A.4 may be part of the possible formatting and processing instructions to be applied for bulked data generation. Additionally, the following parameters may be also included:
    • Periodic bulked data notification: Notifications are sent periodically (e.g. every hour) and may also in a specified time window (e.g. 2AM to 3AM), irrespective of the number of notifications from a Data Source or amount of data that have been bulked. Applicable when the Fetch Flag=false.
    • Feature type is the field defining the type of pre-processing to be applied to the data from the collected event notifications if the bulked data type includes "pre-processed data samples". In this case, for each feature type there is an associated list of events that should be processed according with the feature type. Feature types are: average, maximum, minimum, skewed value from the collected event notification, most frequent value, and least frequent value.
    • Time Window: Specifies the start and stop time for the requested data or analytics.
      • If the Time Window includes a period in the past, then the data or analytics collection is "historical".
      • If the Time Window includes a period in the future, the data or analytics collection is "runtime".
    • (Optional) Minimum and/or maximum number of samples to be included in the bulked data.
    • Fetch flag, when set to false, it indicates that the bulked data needs to be generated and included in notification messages; when set to true, it indicates that bulked data will not be returned in the notification messages and the consumer must fetch the bulk data. The default value of fetch flag is false.
    • Bulked data deadline, which indicates the limit of time for the consumer to fetch bulked data after receiving a notification that the data is available. Applicable when the fetch flag is set to true.
    • Notification Event Clubbing as defined in clause 5A.4, indicates the number of notifications from a data source to be bulked before sending a notification containing the bulked data to the Consumer. Applicable when the Fetch Flag=False.
    • Processing rules: the types of data manipulation to be applied for the requested bulked data and comprises: entity or geographical aggregation level (e.g. per UEs, per AoI), temporal aggregation (i.e. per minute, per hour); anonymization rules (e.g. anonymization of UE identifications).
  • A Notification Target Address (+ Notification Correlation ID), where the Notification Correlation ID is the unique identification for the bulked data being generated for the requesting consumer.
  • ADRF ID or NWDAF ID (or ADRF Set ID or NWDAF Set ID) storing historical data (optional). If known to the consumer, this may be specified to direct a DCCF or an NWDAF to the repository containing historical data.
  • (Optional) ADRF information indicating whether the collected data for the generation of the bulked data are to be stored in an ADRF, and optionally an ADRF ID.
  • (Optional, in case the requested data is Event IDs) Data Source identification to collect the data, e.g. NF Instance (or NF Set) ID from which the data needs to be collected.
The output parameter of the Nnwdaf_DataManagement_Subscribe or Ndccf_DataManagement_Subscribe, or Nadrf_DataManagement_RetrievalSubscribe service operation comprise the subscription correlation ID, which identifies the requested bulked data.
The input parameters of Nnwdaf_DataManagement_Notify or Ndccf_DataManagement_Notify, or Nmfaf_3caDataManagement_Notify, or Nadrf_DataManagement_RetrievalNotify service operation shall contain the Notification Correlation ID, and the generated bulked data when the fetch flag = false. When the fetch flag = true the notifications will contain the Notification Correlation ID, the Fetch Correlation ID and a target address where the generated bulked data may be retrieved. In the case of unsuccessful bulked data generation, the notification will contain an indication of an unsuccessful bulked data generation, optionally with expired bulked data deadline.
The input parameters for the service operation Nnwdaf_DataManagement_Fetch or Ndccf_DataManagement_Fetch, or Nmfaf_3caDataManagement_Fetch, or Nadrf_DataManagement_RetrievalRequest include: the Notification correlation ID (+list of Fetch Correlation ID), which identifies the requested bulked data.
The output parameters for the service operation Nnwdaf_DataManagement_Fetch include:
  • the generated bulked data.
The generated bulked data exposed by the above listed service operations comprises:
  • the dataset (i.e. the resulting set of data samples and/or set of pre-processed data samples from the collected event notifications) generated based on the parameters of bulked data request and instructions for bulked data generation, with each data sample including the following data structure:
    • if the bulked data type is "pre-processed data samples" the data type (e.g. Event ID) and associated feature type (e.g. average values of Event ID) are included;
    • data value (when "data samples" bulk data type is used) or processed values (when "pre-processed data samples" bulked data type is used);
    • timestamp when the data sample is associated with a bulked data.
Up

Up   Top   ToC