Versions Compared

Old Version 2

changes.mady.by.user Ahila P

Saved on

compared with

New Version 3

changes.mady.by.user Ahila P

Saved on

Key

  • This line was added.
  • This line was removed.
  • Formatting was changed.

Refer Intelligent Slicing flow for complete closed loop sequence.

Overview

Image Removed

  1. DCAE - VES Collector receives the file ready events from RAN Simulator and the PM file is fetched from SFTP by Data File Collector (DFC). (Not shown in this picture)
  2. PM Mapper collects this PM file, generates PM events and ML Block receives the PM data from PM mapper.
  3. ML MS before applying the recommendation, it checks with AAI whether the intent is satisfied and the closed loop is triggered.

...

titleSample PM event - Input to ML MS

...

Existing ML MS (not officially exposed to ONAP) requires certain enhancements to satisfy the intent guarantee.

Overview

Image Added


Functionality of ML block is divided into two parts. 

  1. Model Training (Offline)
  2. Validation
  3. Prediction

Step1: ML Offline Training

This step requires a huge amount of historical data for the analysis. Currently, RAN Simulator exposes an API to generate the PM data (xml file) and stored in SFTP. The ML training block can fetch the files from SFTP and use it for training. It requires both the input (PM metrics) and output data (intent). 

draw.io Diagram
bordertrue
diagramNameOffline Training
simpleViewerfalse
width
linksauto
tbstyletop
lboxtrue
diagramWidth811
revision2
Image Removed


PM file format

Expand
titlePM file format for Training

<?xml version="1.0" encoding="UTF-8"?>
<measCollecFile xmlns="http://www.3gpp.org/ftp/specs/archive/32_series/32.435#measCollec">
    <fileHeader dnPrefix="Prefix" fileFormatVersion="32.435 V10.0" vendorName="Acme Ltd">
        <fileSender localDn="cucpserver1"/>
        <measCollec beginTime="2020-10-14T14:39:20.469"/>
    </fileHeader>
    <measData>
        <managedElement localDn="cucpserver1" swVersion="r0.1"/>
        <measInfo measInfoId="measInfoIsVal">
            <job jobId="1118"/>
            <granPeriod duration="PT900S" endTime="2020-10-14T14:39:20.473"/>
            <repPeriod duration="PT900S"/>
            <measType p="1">SM.PDUSessionSetupReq.0011-0010</measType>
            <measType p="2">SM.PDUSessionSetupSucc.0011-0010</measType>
            <measType p="3">SM.PDUSessionSetupFail.0</measType>
            <measType p="4">SM.PDUSessionSetupReq.0010-1110</measType>
            <measType p="5">SM.PDUSessionSetupSucc.0010-1110</measType> 
            <measValue measObjLdn="113025289">
                <r p="4">4364.0</r>
                <r p="5">2739.0</r>
                <r p="3">1517.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="113025290">
                <r p="4">4742.0</r>
                <r p="5">3184.0</r>
                <r p="3">1459.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="113025296">
                <r p="4">5264.0</r>
                <r p="5">3545.0</r>
                <r p="3">1629.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="82268687">
                <r p="4">6952.0</r>
                <r p="5">4337.0</r>
                <r p="3">2363</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="82268689">
                <r p="4">4229.0</r>
                <r p="5">3021.0</r>
                <r p="3">1135.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="95697155">
                <r p="4">4364.0</r>
                <r p="5">3201.0</r>
                <r p="3">1054.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="95697174">
                <r p="4">7041.0</r>
                <r p="5">4229.0</r>
                <r p="3">2599.0</r>
                <suspect>false</suspect>
            </measValue> 
            <measValue measObjLdn="95697175">
                <r p="1">3502.0</r>
                <r p="2">2598.0</r>
                <r p="3">851.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="95697176">
                <r p="1">4858.0</r>
                <r p="2">3430.0</r>
                <r p="3">1295.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="103597825">
                <r p="1">5134.0</r>
                <r p="2">3135.0</r>
                <r p="3">1847.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="103597826">
                <r p="1">4773.0</r>
                <r p="2">3007.0</r>
                <r p="3">1650.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="84327425">
                <r p="1">4573.0</r>
                <r p="2">3347.0</r>
                <r p="3">1111.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="84327426">
                <r p="1">4316.0</r>
                <r p="2">3126.0</r>
                <r p="3">1102.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="103593999">
                <r p="1">5314.0</r>
                <r p="2">3271.0</r>
                <r p="3">1860.0</r>
                <suspect>false</suspect>
            </measValue>
            <measValue measObjLdn="103594000">
                <r p="1">5037.0</r>
                <r p="2">3732.0</r>
                <r p="3">1193.0</r>
                <suspect>false</suspect>
            </measValue>
        </measInfo>
    </measData>
    <fileFooter>
        <measCollec endTime="2020-10-14T14:39:20.474"/>
    </fileFooter>
</measCollecFile>

Impacts:

RAN Simulator:

PM data generation API should include the capability to store it in mongo DB. 

Step2: ML Online Training