Page History

...

Functionality of ML block is divided into two three parts.

Model Training (Offline)
Validation
Prediction

...

draw.io Diagram

border	true

diagramName	Offline Training
simpleViewer	false
width
links	auto
tbstyle	top
lbox	true
diagramWidth	811
revision	23

PM file format

Expand

title	PM 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.

...

Training data format

[
"{\"event\": {\"commonEventHeader\": {\"domain\": \"perf3gpp\",\"eventId\": \"2ff40cb0-377b-49f6-acea-5c7893e53f07\",\"sequence\": 0,\"eventName\": \"perf3gpp_AcmeNode-Acme_pmMeasResult\",\"sourceName\": \"oteNB5309\",\"reportingEntityName\": \"\",\"priority\": \"Normal\",\"startEpochMicrosec\": 1602686360469,\"lastEpochMicrosec\": 1602686360474,\"version\": \"4.0\",\"vesEventListenerVersion\": \"7.1\",\"timeZoneOffset\": \"UTC+05:00\"},\"perf3gppFields\": {\"perf3gppFieldsVersion\": \"1.0\",\"measDataCollection\": {\"granularityPeriod\": 1602686360473,\"measuredEntityUserName\": \"\",\"measuredEntityDn\": \"cucpserver1\",\"measuredEntitySoftwareVersion\": \"r0.1\",\"measInfoList\": [{\"measInfoId\": {\"sMeasInfoId\": \"measInfoIsVal\"},\"measTypes\": {\"sMeasTypesList\":[\"SM.PDUSessionSetupReq.0011-0010\",\"SM.PDUSessionSetupSucc.0011-0010\",\"SM.PDUSessionSetupFail.0\",\"SM.PDUSessionSetupReq.0010-1110\",\"SM.PDUSessionSetupSucc.0010-1110\"]},\"measValuesList\": [{\"measObjInstId\": \"113025289\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"4364\"},{\"p\": 5,\"sValue\": \"2739\"},{\"p\": 3,\"sValue\": \"1517\"}]},{\"measObjInstId\": \"113025290\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"4742\"},{\"p\": 5,\"sValue\": \"3184\"},{\"p\": 3,\"sValue\": \"1459\"}]},{\"measObjInstId\": \"113025296\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"5264\"},{\"p\": 5,\"sValue\": \"3545\"},{\"p\": 3,\"sValue\": \"1629\"}]},{\"measObjInstId\": \"82268687\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"6952\"},{\"p\": 5,\"sValue\": \"4337\"},{\"p\": 3,\"sValue\": \"2363\"}]},{\"measObjInstId\": \"82268689\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"4229\"},{\"p\": 5,\"sValue\": \"3021\"},{\"p\": 3,\"sValue\": \"1135\"}]},{\"measObjInstId\": \"95697155\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"4364\"},{\"p\": 5,\"sValue\": \"3201\"},{\"p\": 3,\"sValue\": \"1054\"}]},{\"measObjInstId\": \"95697174\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 4,\"sValue\": \"7041\"},{\"p\": 5,\"sValue\": \"4229\"},{\"p\": 3,\"sValue\": \"2599\"}]},{\"measObjInstId\": \"95697175\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"3502\"},{\"p\": 2,\"sValue\": \"2598\"},{\"p\": 3,\"sValue\": \"851\"}]},{\"measObjInstId\": \"95697176\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"4858\"},{\"p\": 2,\"sValue\": \"3430\"},{\"p\": 3,\"sValue\": \"1295\"}]},{\"measObjInstId\": \"103597825\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"5134\"},{\"p\": 2,\"sValue\": \"3135\"},{\"p\": 3,\"sValue\": \"1847\"}]},{\"measObjInstId\": \"103597826\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"4773\"},{\"p\": 2,\"sValue\": \"3007\"},{\"p\": 3,\"sValue\": \"1650\"}]},{\"measObjInstId\": \"84327425\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"4573\"},{\"p\": 2,\"sValue\": \"3347\"},{\"p\": 3,\"sValue\": \"1111\"}]},{\"measObjInstId\": \"84327426\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"4316\"},{\"p\": 2,\"sValue\": \"3126\"},{\"p\": 3,\"sValue\": \"1102\"}]},{\"measObjInstId\": \"103593999\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"5314\"},{\"p\": 2,\"sValue\": \"3271\"},{\"p\": 3,\"sValue\": \"1860\"}]},{\"measObjInstId\": \"103594000\",\"suspectFlag\": \"false\",\"measResults\": [{\"p\": 1,\"sValue\": \"5037\"},{\"p\": 2,\"sValue\": \"3732\"},{\"p\": 3,\"sValue\": \"1193\"}]}]}]}}}}"
]

To fetch the Slice Profile (intent) from config DB, below API can be used.

GET: http://localhost:8080/api/sdnc-config-db/v4/profile-config/{nSSAI}

The response will be in the form of:

{

"dLThptPerSlice": 1,

"uLThptPerSlice": 2,

"maxNumberOfConns":300

}

Recommendation for each can be maximum of 5120 and the aggregation of maximumNoOfConnections recommended for a slice (sNSSAI) can be upto 110% (10 % buffer) of the intent.

Step2: Validation

Trained ML model can be validated against the dynamic data in the E2E Slicing set up.

Configure four E2E Slices (shared or non-shared)
Trigger PM data generation for those slices using the below RAN Simulator API

Code Block
curl -X POST -H "Content-Type: text/plain" http://localhost:8081/ransim/api/GenerateIntelligentSlicingPmData -i

3. Run the ML model against the generated PM data and check whether it works as expected

4. When it is not required, stop the pm data generation as it produces huge amount of data that consumes more space.

Code Block
curl -X POST -H "Content-Type: text/plain" http://localhost:8081/ransim/api/stopIntelligentSlicingPmData -i

Step3: Prediction

Trained ML model is now ready to predict and recommend suggestions for the cell level configurations (maxNumberOfConns) for a slice.

Space shortcuts

Page tree

Versions Compared

Old Version 3

New Version 4

Key

Impacts:

Step2: Validation

Step3: Prediction