integration beyond components and models: research challenges and directions

Post on 15-Apr-2017

215 Views

Category:

Software

0 Downloads

Preview:

Click to see full reader

TRANSCRIPT

1

Integration Beyond Components and Models: Directions and Challenges

Ivan Ruchkin

4th Architecture-Centric Virtual Integration WorkshopWICSA/CompArch 2016

Venice, ItalyApril 5, 2016

2

Image cre dit: bankin fosecurity. com

3

Image cre dit: bankin fosecurity. com

Imag

e cr

e dit:

tech

nolo

gyne

wh e

re.w

ordp

ress

.com

4

Image cre dit: bankin fosecurity. com

Imag

e cr

e dit:

tech

nolo

gyne

wh e

re.w

ordp

ress

.com

Image credit: Ajinkya Bhave

5

● Goal: Autonomy in the physical world

6

● Goal: Autonomy in the physical world● But: Heterogeneity of system elements

7

● Goal: Autonomy in the physical world● But: Heterogeneity of system elements● But: Growing complexity and scale

8

● Goal: Autonomy in the physical world● But: Heterogeneity of system elements● But: Growing complexity and scale● Danger: interactions fail → systems fail

9

Integration

Bringing together elements of a system to make them operate cohesively.

Image cre dit: chevin fleet.com

10

Integration

● What have we been doing? – Integration for components; models.

Image cre dit: chevin fleet.com

11

Integration

● What have we been doing? – Integration for components; models.

● What is coming up? – Integration for modeling methods; data; humans.

Image cre dit: chevin fleet.com

12

Component IntegrationIm

age

cre d

it:

tech

nolo

gyne

whe

re. w

ordp

ress

.com

Image credit: Ajinkya Bhave

13

Component IntegrationIm

age

cre d

it:

tech

nolo

gyne

whe

re. w

ordp

ress

.com

Image credit: Ajinkya Bhave

14

Component IntegrationIm

age

cre d

it:

tech

nolo

gyne

whe

re. w

ordp

ress

.com

Image credit: Ajinkya Bhave

15

Component IntegrationIm

age

cre d

it:

tech

nolo

gyne

whe

re. w

ordp

ress

.com

Image credit: Ajinkya Bhave

16

Component IntegrationIm

age

cre d

it:

tech

nolo

gyne

whe

re. w

ordp

ress

.com

Image credit: Ajinkya Bhave

17

Component Integration

● Interface and composition– E.g., FMI [1], automata interfaces [2]

[1] Blochwitz et al. Functional Mockup Interface 2.0: The Standard for Tool independent Exchange of Simulation Models. 2012.[2] Lampka et al. Component-based system design: analytic real-time interfaces for state-based component implementations, STTT 2013.

18

Component Integration

● Interface and composition– E.g., FMI [1], automata interfaces [2]

– Tradeoff: universality vs. tractability

[1] Blochwitz et al. Functional Mockup Interface 2.0: The Standard for Tool independent Exchange of Simulation Models. 2012.[2] Lampka et al. Component-based system design: analytic real-time interfaces for state-based component implementations, STTT 2013.

19

Component Integration

● Interface and composition– E.g., FMI [1], automata interfaces [2]

– Tradeoff: universality vs. tractability

● Compositional reasoning – Contract-based design [3]

[1] Blochwitz et al. Functional Mockup Interface 2.0: The Standard for Tool independent Exchange of Simulation Models. 2012.[2] Lampka et al. Component-based system design: analytic real-time interfaces for state-based component implementations, STTT 2013. [3] Benveniste et al. Contracts for Systems Design: Theory, Research Report, 2015.

20

Component Integration

● Interface and composition– E.g., FMI [1], automata interfaces [2]

– Tradeoff: universality vs. tractability

● Compositional reasoning – Contract-based design [3]

● Shortcoming: cross-cutting quality concerns

[1] Blochwitz et al. Functional Mockup Interface 2.0: The Standard for Tool independent Exchange of Simulation Models. 2012.[2] Lampka et al. Component-based system design: analytic real-time interfaces for state-based component implementations, STTT 2013. [3] Benveniste et al. Contracts for Systems Design: Theory, Research Report, 2015.

21

Model Integration

22

Model Integration

23

Model Integration

1. Abstraction

24

Model Integration

1. Abstraction

2. Relation

25

Model Integration

1. Abstraction

2. Relation

Structural Behavioral

...

26

● On the structural side:– Metamodel composition [4]

– Architectural views [5]

Model Integration

[4] Passarini et al. Cyber-physical systems design: transition from functional to architectural models, DAES 2015. [5] Bhave et al. View Consistency in Architectures for Cyber-Physical Systems, ICCPS 2011.

27

● On the structural side:– Metamodel composition [4]

– Architectural views [5]

● On the behavioral side: – Heterogeneous simulation [6]

– Behavior relations [7]

Model Integration

[4] Passarini et al. Cyber-physical systems design: transition from functional to architectural models, DAES 2015. [5] Bhave et al. View Consistency in Architectures for Cyber-Physical Systems, ICCPS 2011. [6] Eker et al. Taming heterogeneity - the Ptolemy approach, Proc. of IEEE 20013. [7] Rajhans et al. Supporting Heterogeneity in Cyber-Physical Systems Architectures, TAC 2014.

28

● On the structural side:– Metamodel composition [4]

– Architectural views [5]

● On the behavioral side: – Heterogeneous simulation [6]

– Behavior relations [7]

● Shortcoming: fragility in the face of change

Model Integration

[4] Passarini et al. Cyber-physical systems design: transition from functional to architectural models, DAES 2015. [5] Bhave et al. View Consistency in Architectures for Cyber-Physical Systems, ICCPS 2011. [6] Eker et al. Taming heterogeneity - the Ptolemy approach, Proc. of IEEE 20013. [7] Rajhans et al. Supporting Heterogeneity in Cyber-Physical Systems Architectures, TAC 2014.

29

Integration

● What have we been doing?

– Integration for components; models.● What is coming up?

– Integration for modeling methods; data; humans.

Image cre dit: chevin fleet.com

30

Modeling Method Integration

31

Modeling Method Integration

32

Modeling Method Integration

Focus: analysis/transformation procedures

33

Modeling Method Integration

● Techniques: – Dependency management [8]

[8] A. Qamar. Model and Dependency Management in Mechatronic Design, PhD Thesis, KTH 2013.

34

Modeling Method Integration

● Techniques: – Dependency management [8]

– Assumption verification [9]

[8] A. Qamar. Model and Dependency Management in Mechatronic Design, PhD Thesis, KTH 2013.[9] Ruchkin et al. Contract-based Integration of Cyber-physical Analyses, EMSOFT 2014.

35

Modeling Method Integration

● Techniques: – Dependency management [8]

– Assumption verification [9]

● How can evolution of sets of heterogeneous CPS models be systematically supported?

[8] A. Qamar. Model and Dependency Management in Mechatronic Design, PhD Thesis, KTH 2013.[9] Ruchkin et al. Contract-based Integration of Cyber-physical Analyses, EMSOFT 2014.

36

Modeling Method Integration

● Techniques: – Dependency management [8]

– Assumption verification [9]

● How can evolution of sets of heterogeneous CPS models be systematically supported?

● How can tools, processes, and methods for CPS modeling be integrated?

[8] A. Qamar. Model and Dependency Management in Mechatronic Design, PhD Thesis, KTH 2013.[9] Ruchkin et al. Contract-based Integration of Cyber-physical Analyses, EMSOFT 2014.

37

Data Integration

38

Data IntegrationIm

age cre dit: Deshe ng Z

hang

39

Data Integration

Focus: heterogeneous datasets from CPS elements

Image cre dit: D

eshe ng Zhang

40

Data Integration

● How can data incompleteness in CPS design be detected and compensated for?

41

Data Integration

● How can data incompleteness in CPS design be detected and compensated for?

● How can model-based and data-centric approaches to system design be (non-trivially) synergized?

42

Integration with Humans

● Humans as external agents – “Human-in-the-loop”

43

Integration with Humans

● Humans as external agents – “Human-in-the-loop”

Image cre dit: historia viation.co m

44

Integration with Humans

● Humans as external agents – “Human-in-the-loop”

● How can humans be given adequate comprehension and control of complex systems?

45

Integration with Humans

● Humans as external agents – “Human-in-the-loop”

● How can humans be given adequate comprehension and control of complex systems?

● How can competing theories of human cognition be reconciled in practical human models?

46

Integration with Humans

● Humans as external agents – “Human-in-the-loop”

● How can humans be given adequate comprehension and control of complex systems?

● How can competing theories of human cognition be reconciled in practical human models?

● How can contextual fragility of human models be bridged?

47

Integration with Humans

● Humans as engineers

48

Integration with Humans

Computer Science Electrical Engineering

Mechanical Engineering

● Humans as engineers

49

Integration with Humans

● Humans as engineers ● How do the inherent biases of each CPS

discipline affect design and development?

50

Integration with Humans

● Humans as engineers ● How do the inherent biases of each CPS

discipline affect design and development? ● What are the shared concepts, conflicts, and

omissions at the boundaries of disciplines?

51

Summary

● In CPS integration overcomes heterogeneity and complexity.● Foundations of integration:

– Components

– Models

● Emerging directions of integration: – Modeling methods

– Data

– Humans

● Takeaway: let's broaden the horizons of integration!

top related