Top Banner
Dealing with App validation, in practice 5GPPP-TB eWorkshop 2020-12-10 Manuel Lorenzo (Ericsson Spain, 5G EVE & 5GROWTH) Aitor Zabala (Telcaria Ideas, 5GROWTH) www.5g-ppp.eu 1 2020-12-10
38

5GPPP-TB eWorkshop

Oct 05, 2021

Download

Documents

dariahiddleston
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 5GPPP-TB eWorkshop

Dealing with App validation, in practice5GPPP-TB eWorkshop

2020-12-10Manuel Lorenzo (Ericsson Spain, 5G EVE & 5GROWTH)

Aitor Zabala (Telcaria Ideas, 5GROWTH)

www.5g-ppp.eu 12020-12-10

Page 2: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-12-10 www.5g-ppp.eu 2

Page 3: 5GPPP-TB eWorkshop

The Challenge

2020-12-10 www.5g-ppp.eu 3

Application

ApplicationDeployment

Model

ApplicationTraffic Model

ApplicationKPIs

Network

Technology & Configuration

Choices

Network Environment& Conditions

NetworkKPIs

• Compatible

• Functional

• Performing

Application Validatedover 5G Networks

Page 4: 5GPPP-TB eWorkshop

The Challenge

How to validate vertical-specific KPIs over a 5G validation platform• Understanding the relative influence of 5G network

performance indicators• Collecting and processing performance data collected

in the actual experimentation• Iterating the validation process towards the

optimization of both application and network performance

2020-12-10 www.5g-ppp.eu 4

Page 5: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-09-15 www.5g-ppp.eu 5

Page 6: 5GPPP-TB eWorkshop

Definitions

Vertical KPIs

• Targeted values for a vertical-specific KPI. Normally influenced –although normally not only– by one or several 5G service KPIs.

• Specific to the UC, in value and sometimes in nature.

• Measured by the application.• It’s NOT part of the SLA between the

vertical and the CSP. It’s only meaningful in the specific vertical domain

• Ex: Device to Controller synchronization = 99% of device-controller time-outs do no expire for a 1h-long traffic session

2020-12-10 www.5g-ppp.eu 6

5G Core KPIs5G Network KPIs:• Network Performance• Measured in the Network (with network testing tools)• Use case independent.• Part of the SLA between the 5G CSP and the 5G

technology/solution vendor• Ex: For instance U/L Latency OWD = 4 ms (avg), measured

with standard network monitoring tools (active probe)5G Service KPIs:• Perceived/experienced Network performance.• Measured by the vertical (itself or by trusted platform).• Values are specific to UCs (i.e may vary from UC to UC)• Part of the SLA between the Vertical and the 5G CSP• Note: Let’s consider, just for a while, only the type of 5G

service KPIs related directly to a single reference 5G network KPI (for simplicity)

• Ex: U/L Latency for the actual application TCP traffic = 8 ms (avg), measured by trusted ICT17 (with passive probe)

Page 7: 5GPPP-TB eWorkshop

Relevant Clarification

www.5g-ppp.eu 72020-12-10

5G NETWORK KPINETWORK PERFORMANCE

(PRECONDITION)EXPER. PERFORMANCE

(POST-CONDITION)5G SERVICE KPI

E2E LATENCY YES Not likely -

RTT LATENCY YES YESMAXIMUM EXPECTED

RTT LATENCY

USER DATA RATE (D/L or U/L) YES YESMINIMUM EXPECTED

USER DATA RATE (D/L or U/L)

PEAK DATA RATE (D/L or U/L) YES Not Applicable -

RELIABILITY YES YESMINIMUM EXPECTED

RELIABILITY FOR A SELECTED MAX RTT LATENCY VALUE

AVAILABILITY (NETWORK) YES Not Applicable -

AVAILABILITY (SERVICE) Not Applicable YESMINIMUM EXPECTED

AVAILABILITY (SERVICE)

NETWORK PERFORMANCE DIMENSION SERVICE EXPERIENCE DIMENSION

Page 8: 5GPPP-TB eWorkshop

Reference Framework (by 2018)

2020-12-10 www.5g-ppp.eu 8

KPI1

KPI2KPI8

KPI4KPI6

KPI3KPI7

KPI5

Ideal5GLTE

KPI1

KPI2KPI8

KPI4KPI6

KPI3KPI7

KPI5

5G KPIs demanded by the Vertical App / Use CsseExpected Network Performance levels per Technology

Page 9: 5GPPP-TB eWorkshop

Reference Framework (by 2018)

VERTICAL APPLICATION

5G Core KPI-1 5G Core KPI-2 5G Core KPI-3 …

UC-1 -

UC-2 -

UC-3

2020-12-10 www.5g-ppp.eu 9

1. Vertical Application is broken down into a set of Use Cases2. Each Use Case poses quantitative requirements on 5G Core KPIs3. Validation activities focus on the set of 5G Core KPIsNote: The performance of the Application itself is neither formally modelled nor monitored

Page 10: 5GPPP-TB eWorkshop

Reference Framework (2019-2020)

USECASE

5G Core KPI-1 5G Core KPI-2 5G Core KPI-3 …

Composite KPI-A - -

Composite KPI-B - -

Composite KPI-C -

2020-12-10 www.5g-ppp.eu 10

1. Vertical Application is broken down into a set of Use Cases2. For each Use Case a set of Composite KPIs are modelled over component 5G Core KPIs,

and quantitiative requirements are posed on the 5G Core KPIs3. Validation activities focus on the set of component 5G Core KPIsNote: Modelling & Monitoring of composite KPIs is a major step towards formal app validation

Page 11: 5GPPP-TB eWorkshop

Reference Framework (twds 2021)

VERTICAL APPLICATION KPI

5G SERVICE KPI-1 5G SERVICE KPI-2 5G SERVICE KPI-3 …

V.KPI-A = <Value> -

V.KPI-B = <Value> -

V.KPI-C = <Value> - - -

2020-12-10 www.5g-ppp.eu 11

1. Vertical Application performance is modelled, in its own domain, in meaningful, quantifiableand monitorable Vertical KPIs

2. For each V-KPI a set of -a priori- influencing 5G Core KPIs are postulated3. Validation activities focus on BOTH V-KPI validation and 5G Core KPI Validation4. Post-validation data analysis can bring new insight on the actual V.KPI-vs-C.KPI model

Page 12: 5GPPP-TB eWorkshop

Visualizing the model …

2020-12-10 www.5g-ppp.eu 12

• For simplicity, let’s limit now to 2 just dimensions per V-KPI

Page 13: 5GPPP-TB eWorkshop

Technology & Limits

2020-12-10 www.5g-ppp.eu 13

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Beyond currenttechnologypossibilities

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

Page 14: 5GPPP-TB eWorkshop

This is an a priori V.KPI model

2020-12-10 www.5g-ppp.eu 14

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

5G Service KPI1

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 15: 5GPPP-TB eWorkshop

Data Collection Scenarios

2020-12-10 www.5g-ppp.eu 15

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 16: 5GPPP-TB eWorkshop

By collecting vertical KPI data …

2020-12-10 www.5g-ppp.eu 16

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 17: 5GPPP-TB eWorkshop

.. we can tune our a priori model

2020-12-10 www.5g-ppp.eu 17

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 18: 5GPPP-TB eWorkshop

Buit it could be worse!

2020-12-10 www.5g-ppp.eu 18

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 19: 5GPPP-TB eWorkshop

It could even be inconsistent!

2020-12-10 www.5g-ppp.eu 19

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 20: 5GPPP-TB eWorkshop

Reference Framework (twds 2021)

VERTICAL APPLICATION KPI

5G SERVICE KPI-1 5G SERVICE KPI-2 5G SERVICE KPI-3 …

V.KPI-A = <Value> -

V.KPI-B = <Value> -

V.KPI-C = <Value> - - -

2020-12-10 www.5g-ppp.eu 20

• It’s all about data analysis mixing Both V-KPIs and C.KPIs involved in the experimentation• The actual V.KPI-vs-C.KPI model is an OUTPUT of the validation process (rather than an input)• The range of 5G Core KPIs supporting the V-KPI expected levels allows to select the 5G

network technology and configurations enabling it.

Page 21: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-12-10 www.5g-ppp.eu 21

Page 22: 5GPPP-TB eWorkshop

Anything else?

• Yes,– Additionally we could tune, if deemed necessary towards

further eventual tests,• Mobile Network

– Technology– Configurations

• Vertical SW application– Design– Deployment model

– and re-start the cycle

2020-12-10 www.5g-ppp.eu 22

Page 23: 5GPPP-TB eWorkshop

Tuning of NW configuration

2020-12-10 www.5g-ppp.eu 23

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 24: 5GPPP-TB eWorkshop

Tuning of Vertical application

2020-12-10 www.5g-ppp.eu 24

KPI1

(ex: Reliability)

KPI2

(ex: Latency) Theoretical Best(ex: 1,0 ms)

Theoretical Best(ex: 99,999%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 99,900%)

Former Best(ex: 15 ms)

2

5G Service KPI1

1

3 4

5G

Ser

vie

KP

I 2

Beyond currenttechnologypossibilities

Page 25: 5GPPP-TB eWorkshop

What does it take?

2020-12-10 www.5g-ppp.eu 25

Page 26: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-12-10 www.5g-ppp.eu 26

Page 27: 5GPPP-TB eWorkshop

Application Case – 5G EVE

2020-12-10 www.5g-ppp.eu 27

Page 28: 5GPPP-TB eWorkshop

Monitoring Latency

www.5g-ppp.eu 282020-12-10

LATENCY

Probe

ProbeProbe

OWD

E2E LATENCY

RTT

Page 29: 5GPPP-TB eWorkshop

Monitoring Data Rate

www.5g-ppp.eu 292020-12-10

U/L & D/L DATA RATES

Probe

ProbeProbe

D/L DATA RATE U/L DATA RATE

U/L & D/L DATA RATES

Page 30: 5GPPP-TB eWorkshop

Case 3GPP Standards Access Core Application 5G KPIs

3GPP Release

3GPP Architecture option

Band BW(MHz)

Pattern MIMO Layers

Modulation Aggregation Deployment Deployment UP Latency(ms)

Reliability(%)

UL Peak/ User Data

Rate (Mbps)

• Rel14

• Rel15

• Rel16

• Legacy

• NSA• SA

• LB• M

B• HB

• 20• 50,• 100,

200,…

• FDD• 4:1• 7:3

• 1• 2• 4• (8)

• 256QAM(DL)

• 64QAM(UL)

• Single Carrier

• Carrier Aggreg.

• NR+LTE

• Edge• Central

• Local Service

• Central Service

• OTT

Exp#1:2020-09-21

UC-1

Rel 15 NSA MB 50MHz • 7:3 4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local Service(Edge)

• 3.8msOWD E2E ICMP

• 7.5 RTT

99%(RTT=9ms)

Exp#1:2020-09-21

UC-1

Rel 15 NSA MB 50MHz • 7:3 4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Central Service

• Regional

+2ms• National

+4ms

E2E• 4.1ms

OWD ICMP

• 5.2msOWDICMP

99%• RTT=10.5ms• RTT=12ms

Exp#2:2020-10-05

UC-1

Rel 15 NSA MB 50MHz • 7:3 4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local Service (Edge)

• 22msE2E OWD

TCP UserTraffic

99%(RTT=32ms)

• 40MbpsTCP HD video streaming

Exp#2:2020-10-05

UC-1

Rel 15 NSA MB 50MHz • 7:3 4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Central Service

• Regional

+2ms• National

+4ms

• 26msE2E OWD

TCP UserTraffic• 27ms

99%• RTT=34ms• RTT=36ms

Exp#3:2020-10-05

UC-1

Rel 15 NSA MB 50MHz • 7:3 4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local Service(Edge ->

Application)

• 54MbpsIperf3 UDP

5G EVEI4.0 +

Computer

Vision

@EdgeUse Case

Page 31: 5GPPP-TB eWorkshop

Case 3GPP Standards Access Core Application 5G KPIs

3GPP Release

3GPP Architecture

optionBand

BW(MHz)

Pattern MIMO Layers Modulation Aggregation Deployment Deployment

UP Latency(ms)E2E

DL Peak/ UserData Rate

(Mbps)

UL Peak/ UserData Rate

(Mbps)• Rel 14• Rel 15• Rel 16

• Legacy• NSA• SA

• LB• MB• HB

• 20• 50,• 100,• 200,

• FDD• 4:1• 7:3

• 1• 2• 4• (8)

• 256QAM(DL)

• 64QAM(UL)

• Single Carrier

• Carrier Aggreg.

• NR+LTE

• Edge• Central

• Local Service

• Central Service

• OTT

Rel 14 Legacy MB 20MHz • FDD2 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local• 12ms• 12ms

• 195 MbpsTCP

• 75 MbpsTCP

Rel 15 NSA MB 50MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local• 6ms• 5ms

• 776 Mbps• 660 MbpsTCP

• 41 Mbps• 63 MbpsTCP

Rel 15 NSA MB 100MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local• 6ms• 5ms

• 1600 Mbps• 1350 MbpsTCP

• 86 Mbps• 130 MbpsTCP

Rel 15 NSA MB 50MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

DownlinkNR (50MHz) +LTE (20MHz)

Edge Local• 12ms• 11ms

• 950 Mbps• 810 MbpsTCP

• 41 Mbps• 63 MbpsTCP

Rel 15 NSA MB 50MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

UplinkNR (50MHz) +LTE (20MHz)

Edge Local• 12ms• 11ms

• 776 Mbps• 660 MbpsTCP

• 100Mbps• 120MbpsTCP

Rel 15 NSA MB 50MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Downl Carrier Aggregation

(50MHz + 20MHz)

Edge Local• 7ms• 6ms

• 1050 Mbps• 860 MbpsTCP

• 41 Mbps• 63 MbpsTCP

Rel 15 NSA HB 100MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local • 3ms• 1470 Mbps• 1160 MbpsTCP

• 50 Mbps• 82 MbpsTCP

Rel 15 NSA HB 400MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Single Carrier Edge Local • 3ms• 5900 Mbps• 4630 MbpsTCP

• 200Mbps• 330MbpsTCP

Rel 15 NSA HB 100MHz• 4:1• 7:3

4 DL /1 UL

256QAM DL / 64QAM UL

Downl Carrier Aggregation(100MHz + 100MHz + 100MHz + 100MHz)

Edge Local • 3ms• 5800 Mbps• 4560 MbpsTCP

• 180Mbps• 300MbpsTCP

MAX THEORET

-ICALKPI

VALUES

Page 32: 5GPPP-TB eWorkshop

Practical best (ICMP)

Real deployment results

2020-12-10 www.5g-ppp.eu 32

KPI1

(ex: Reliability)

KPI2

(ex: OWD Latency)

Theoretical Best(ex: 99,9%)

Ideal(100%)

Ideal(0 ms)

Satisfied byformer

technologies

Former Best(ex: 98%)

Former Best(ex: 40 ms)

5G Service KPI1

5G

Ser

vice

KP

I 2

Rel 14 Legacy – CS Regional

Rel 15 NSA – CS Regional

Rel 15 NSA – Local Service

5G EVEI4.0 + Computer Vision

@Edge Use Case

3,2ms4,1ms14ms

32ms

26ms

22ms

Rel14 Theor. Best

(12 ms)

Rel15 Theor. Best

(3 ms)

99%

Beyond currenttechnologypossibilities

Page 33: 5GPPP-TB eWorkshop

Application case (5GROWTH – Innovalia )

2020-12-10 www.5g-ppp.eu 33

5 ms

5 ms

5 ms

5 ms

0 ms

0 ms

?

?

+ 0.5 ms per 100 km

+ 0.5 ms per 100 km

KPIs

Distance

KPI Total Delay(excl. camera)

S-KPI Synch

20 ms

(50 ms)

95% 90%

40%

KPI Reliability 99,9% 99,9% 99,9%

Distance = 0 Distance = 500 km Distance = 1000 km

25 ms30 ms

5 msKPI OWD 5 ms 5 ms

Page 34: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-12-10 www.5g-ppp.eu 34

Page 35: 5GPPP-TB eWorkshop

5G-EVE & 5GROWTH common approach

• Common platform to ship, collect, store and visualize the data generated from the experiments.

• Data can be collected at any domain, even inside the vertical applications.

• Seamlessly adapting to the KPI validation framework defined.

2020-12-10 www.5g-ppp.eu 35

Page 36: 5GPPP-TB eWorkshop

Outline

• The Challenge

• Reference Framework for KPI Modelling & Validation

• KPI Validation ++: an iterative process

• Experimentation Cases

• Approach of 5G EVE (ICT-17) + 5GROWTH (ICT-19)

• Key Take Aways

2020-12-10 www.5g-ppp.eu 36

Page 37: 5GPPP-TB eWorkshop

Key Take-Aways

• KPI models are instrumental for proper and formal application validation procedures

• Vertical KPIs can bring objectivity to the validation and support data-driven analysis and decisions

• Iterating KPI models, network technology and configuration choices, and application design and deployment model can prove a fruitful approach

2020-12-10 www.5g-ppp.eu 37

Page 38: 5GPPP-TB eWorkshop

Thank you!

• Manuel Lorenzo [email protected]

• Aitor [email protected]

2020-12-10 www.5g-ppp.eu 38