core-ncsrd / .github

0 stars 0 forks source link

NWDAF APIs and exposure for the AI/ML models trained on Amarisoft data #9

Open mchristopoulou opened 1 week ago

mchristopoulou commented 1 week ago

Discussed in https://github.com/orgs/core-ncsrd/discussions/3

Originally posted by **xilouris** September 26, 2024 # Team Members - @mchristopoulou - @xilouris - @TheReaperGR - @AthinaVek # Tasks Following our discussions, @TheReaperGR will implement the API to support the output analytics for the use case "Abnormal UE behavior detection" from TS 23.288 https://www.etsi.org/deliver/etsi_ts/123200_123299/123288/17.04.00_60/ts_123288v170400p.pdf > _Corresponding to the "abnormal behaviour" Analytics ID, the analytics result provided by the NWDAF is defined in > Table 6.7.5.3-1 and Table 6.7.5.3-2. When the level of an exception trespasses above or below the threshold, the > NWDAF shall notify the consumer with the exception ID associated with the exception if the exception ID is within the > list of exception IDs indicated by the consumer or matches the expected analytics type indicated by the consumer. The > NWDAF shall provide the Exception Level and determine which of the other information elements to provide, > depending on the observed exception._ ![image](https://github.com/user-attachments/assets/3235fbaa-c473-4e66-826b-161be70614ab) ![image](https://github.com/user-attachments/assets/94160aef-cbfb-4ff5-a7bb-84dcf0496d3f) Therefore, @TheReaperGR will work on the development of the API side and @AthinaVek will provide the available output information from the AI/ML models she has been working on (select the model with the best performance). ### Targeted Output Analytics - Exception ID - Exception Level - Exception trend - optional, future enhancement - UE characteristics (TAC) - SUPI list (use imeisv) - Ratio - Amount - Confidence General information regarding NWDAF are found below. # Objective Develop the function and the APIs that will allow Amarisoft collected data to be available to other internal or external NFs of the 5G Core. # Definitions NWDAF is designed to overcome market fragmentation and proprietary solutions in the area of network analytics, streamlining the way core network data is produced and consumed, as well as generating insights and taking actions based on these insights. NWDAF addresses three primary standardization points: - Data collection interface from network nodes - Predefined analytics insights - Data exposure interface for consumers # Resources From the 3GPP: - [TS 29520 - Network Data Analytics Services](https://www.etsi.org/deliver/etsi_ts/129500_129599/129520/16.04.00_60/ts_129520v160400p.pdf) Table 5.1.6.1-1 specifies the data types defined for the Nnwdaf_EventsSubscription service based interface protocol. - [TS29520_Nnwdaf_AnalyticsInfo.yaml](https://www.3gpp.org/ftp/Specs/archive/OpenAPI/Rel-16/TS29520_Nnwdaf_AnalyticsInfo.yaml) - [TS29520_Nnwdaf_EventsSubscription.yaml](https://www.3gpp.org/ftp/Specs/archive/OpenAPI/Rel-16/TS29520_Nnwdaf_EventsSubscription.yaml)
mchristopoulou commented 1 week ago

@xilouris when time allows, can you check we can add it to privateer-backlog, it was not available as an option. I added it to CORE backlog.

TheReaperGR commented 1 week ago

Me and Athina have successfully produced a json file attached below with data from the ML model and i have created a route in the API so the data can be stored and pulled from the influxDB screenshots attached Screenshot from 2024-10-14 15-19-19 Screenshot from 2024-10-14 15-19-28 Screenshot from 2024-10-14 15-21-17 sample.json

mchristopoulou commented 1 week ago

Looks good! We will meet up during this week to see the demo and decide the next steps.