Paper presentation at SAM'21 conference.

1. 1
Iván Alfonso1,2, Kelly Garcés1,
Harold Castro1, Jordi Cabot2,3
Modeling self-adaptative
IoT architectures
1 University of the Andes, Bogotá, Colombia
2 Open University of Catalonia, Barcelona, Spain
3 ICREA, Barcelona, Spain
October 2021

2. Internet of Things 2
There will be 5.3 billion
internet users in 2023
(Cisco Annual Internet
Report)

3. Cloud
IoT Architecture 3
Network
device
Network
device
Network
device
Cloud
Device Layer
Collecting data
Sensors and actuators
Cloud Layer
Data processing and analysis
Business logic
Data storage
Data centers
Network
device
Network
device
Network
device
Cloud
Fog
node
Fog
node
Cloud Layer
Big data processing
Business logic
Data warehousing
Fog Layer
Local network
Local IoT analytics
Local storage
Edge Layer
Real time data processing
Micro sata storage
Low latency
Device Layer
Collecting data
Sensors and actuators
Cloud-based Multi-layer

4. Dynamic IoT Systems
• Unstable signal strength
• Latency growth
• Software failures
4
• Auto-scaling
• Offloading tasks
• Redeployment services
Unexpected events at runtime Adaptations

5. Example Scenario 5
Coal

6. Example Scenario 6

7. Example Scenario 7

8. Example Scenario 8

9. Example Scenario 9

10. Domain Specific Language (DSL) 10
{
__;
_____;
}
DSL
• Modeling (four layers) of an IoT system,
including IoT devices, edge, fog, and cloud
nodes.
• Modeling the deployment of container-based
applications.
• A specific sublanguage to express adaptation

11. Modeling of the IoT Architecture - Metamodel 11

12. Modeling of the IoT Architecture - Metamodel 12

13. Modeling of the IoT Architecture - Metamodel 13

14. Modeling of the IoT Architecture - Metamodel 14

15. Modeling of the IoT Architecture - Metamodel 15

16. Modeling of the IoT Architecture - Metamodel 16

17. Concrete Syntax 17
Tree view Textual Tabular
{
____;{
_______;
}
_____;
}
• Regions • Applications
• Nodes
• Containers
• IoT devices
• Clusters
• Nodes
• IoT devices

18. Tree View Notation 18
Region Modeling

19. Textual and tabular notation 19
IoT Device Modeling (Textual)

20. Textual and tabular notation 20
IoT Device Modeling (Textual)
IoT Devices Modeling (Tabular)

21. Textual and tabular notation 21
Node Modeling (Textual)
Node Modeling (Tabular)

22. Modeling of the Self-Adaptation Rules 22

23. Modeling of the Self-Adaptation Rules 23

24. Modeling of the Self-Adaptation Rules 24

25. Modeling of the Self-Adaptation Rules 25

26. Self-Adaptation Rule 26
Adaptation Rule (EdgeWF1 node overloaded)

27. Self-Adaptation Rule 27
Adaptation Rule (EdgeWF1 node overloaded)

28. Self-Adaptation Rule 28
Adaptation Rule (EdgeWF1 node overloaded)

29. Self-Adaptation Rule 29
Adaptation Rule (EdgeWF1 node overloaded)

30. Self-Adaptation Rule 30
Adaptation Rule (high gas concentration)

31. Tool Support 31
Model
Code
Generator
YAML
file
IoT System
MPS

32. Conclusion
& Future
Work
• DSL for modeling IoT systems
• Multi-layered architectures
• Self-adaptation rules
• First steps towards a self-adaptive IoT
system architecture
• Extend the DSL to express more complex
adaptation strategies
• Extend the code generator to cover the dynamic
aspects
• Validate the DSL in the mining industry
FUTURE

33. 33
Iván Alfonso1,2, Kelly Garcés1,
Harold Castro1, Jordi Cabot2,3
Thanks
1 University of the Andes, Bogotá, Colombia
2 Open University of Catalonia, Barcelona, Spain
3 ICREA, Barcelona, Spain
October 2021

34. Repository
https://github.com/SOM-Research/selfadaptive-IoT-DSL.git
34

35. Future Approach 35
Developer
IoT System
IoT architecture
(YAML)
Adaptation rules
(PromQL)
DESIGN TIME
Prometheus
RUN TIME
Orchestrator
Model
Code
Generator
Adapter
Alert
Manager