A Practical Application of Monitoring data and AI for ...
Transcript of A Practical Application of Monitoring data and AI for ...
A Practical Application of Monitoring data and AI for Network Management in an Industrial Environment
July 2021Diego R. López – TelefónicaIgnacio Domínguez, Daniel González - UPM
Slice B
Slice C
Slice A
Data center (hosting cloud)
Edge data center(hosting edge cloud)
5G End-to-End Infrastructure(Radio, Transport, Cloud)
Orchestrator
Radio
Industrial Applications
5G-Enabled Growth in Vertical Industries
2
Orchestration
Mission of the Project: Automatically build, deploy and managedifferent vertical industrial services over a shared End-to-End 5Gsystem (incl. Application Domain, Infrastructure Domain, OrchestrationDomain), and perform real field trials validation and demonstrations
• 5Growth aims to perform real field trialsinvolving customer sites of four vertical locations in Portugal, Spain and Italy
• This requires the development, installation, validation and testing of commercial 5G radio, transport and core technologies in vertical sites, connected with the ICT-17 platforms
• PilotsIndustry 4.0• INNOVALIA
• COMAU
Energy• EFACEC_E
Transportation• EFACEC_S
Pilots
3Energy
Transportation
Vertical premises
Vertical premises
5G EVE
Industry 4.0 Industry 4.0
5Growth Innovation Focus
4
Vertical Support
Monitoring Orchestration
Data Infrastructure
Control & Management (Closed-loop, AI/ML)
End-to-end orchestration
Smart Orchestration &
Control
Anomaly Detection
Forecasting and Inference
Security
CI/CD
Containerization
ü Smartness ü Flexibility ü Efficiency ü Automationü Performanceü Security Framework Architecture Algorithms
Data Infrastructre
The Three-Dimensional Closed-Loop
5
Closed-loops for the processes of
• Collecting monitoring data from the services and networks
• Performing real-time data analytics for identifying events to handle
• Making orchestration decisions for optimization and/or reconfiguration of the system.
Digital Twin appInterface
Coordinates
Commands
EDGENETWORK
11
Drivers
Motion planningControl
Use Case: Robotic Digital Twin
RAN
VNF 2 target latency: 60 ms
VNF 1target latency: 600 ms
6
VNF
5Gr-RL
5Gr-VS
WIM
5Gr-SO
Mon
itorin
g
5Growth MANO and Monitoring
Model Execution
BarcelonaTorino Scalingoperation
Deployed Scenario
Heterogeneous WAN transport network
VNF VNF
VIM
Scalingoperation
VPN (data, control and management)
AI/ML Platform
YARN cluster manager
Spark
Interface Manager
BigDLRay
MLlib
AI/ML model register
AI/ML param. register
HDFS dataset storage
ML lifecycle Mng
Interface for model onboarding
Monitoring jobs configuration
1
3
AIML model upload and training
5Gr-Service Orchestrator 5Gr-VoMS 5Gr-AI/ML Platform5Gr-VS
Service request2
Deployment and Configuration
Dataflow configuration4
Real-time performance metrics5
AIML model request/response6
Data consumption7
Scaling workflow
Model execution and scaling triggering8
3 Monitoring jobs configuration
Topic configuration4
5 7 8
• 5Growth addresses data-driven network management in a three-dimensional closed loop• The common double loop for learning and inference• An additional control loop for monitoring data flows
• Making use of the project data infrastructure• Data-driven requirements embedded in service descriptors• Monitoring orchestration• Model-based data flows
• Applied to a real industrial use case• A path to be explored for other classes of Digital Twins
Concluding
11
• J. Baranda, J. Mangues, E. Zeydan, L. Vettori, R. Martínez, X. Li, A. Garcia-Saavedra, C. Fabiana Chiasserini, C. Casetti, K. Tomakh, O. Kolodiazhnyi, C. Jesus Bernardos: On the Integration of AI/ML-based scaling operations in the 5Growth platform, in Proceedings of the 6th IEEE Conference on Network Functions Virtualization and Software Defined Networking (IEEE NFV-SDN 2020), 9-12 November 2020.
• J. Baranda, J. Mangues, E. Zeydan, C. Casetti, C. Fabiana Chiasserini, M. Malinverno, C. Puligheddu, M. Groshev, C. Guimarães, K. Tomakh, D. Kucherenko, O. Kolodiaznnyi, Demo: AIML-as-a-Service for SLA management of a Digital Twin Virtual Network Service , in Proceedings of the International Conference on Computer Communications (IEEE INFOCOM), 10-13 May 2021.
• The whole use case demo on YouTube:• https://youtu.be/7V3AKSrWzzY (short)• https://youtu.be/K5GyrAD7h_Q (long)
More Can Be Found at…
12