+ All Categories
Home > Documents > AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2...

AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2...

Date post: 28-Sep-2020
Category:
Upload: others
View: 0 times
Download: 0 times
Share this document with a friend
26
- ALBERT-LUDWIGS UNIVERSITÄT FREIBURG AVACS Automatic Verification and Analysis of Complex Systems Werner Damm AVACS coordinator
Transcript
Page 1: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

AVACS Automatic Verification and Analysis

of Complex Systems

Werner Damm

AVACS coordinator

Page 2: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

2

Structure of Presentation

The AVACS Vision

Highlights of Phase II

Page 3: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

3

Copyright Prevent Project

Complex Systems

Source: Aramis Project

Page 4: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

4

The Application Context

• Complex Embedded Systems are key enablers for safe flight and safe ground transportation

• Exponential growth in system complexity is a challenge for quality assurance

• AVACS contributes to meeting forthcoming requirements of pertinent safety standards on use of formal analysis methods

• Methods and tools cover large class of “cyber physical systems” seen to be highly relevant for addressing societal challenges (health, security, green mobility, …)

Page 5: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

The AVACS

model- and

requirement space

Page 6: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

6

To Cover the Model- and Requirement Space of Complex Safety Critical Systems

with Automatic Verification Methods

Giving Mathematical Evidence

of Compliance of Models

To Dependability, Coordination, Control

and Real-Time Requirements

The AVACS Vision

Page 7: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

Research

Areas

R

Real-Time

H

Hybrid

S

Coarse

Grain System

Structure

7

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Page 8: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

Sites and PIs Systems & Models

• Domain Expertise

• Specification and verification of

aembedded systems

• Control, Real-Time, Hybrid

• SAT(T)

Algorithmic Aspects

• Decision Procedures

• Constraint solving

• ILP

• Probabilistic model checking

• Abstract Interpretation

• Shape Analysis

Analysis of Extremely

Large State Spaces

• Heuristic Planning

• SAT, BDD, AIG

• Directed Model Checking

• Abstraction

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Ernst Althaus

Bernd Finkbeiner

Sebastian Hack

Holger Herrmanns

Jan Reinecke

Werner Damm

Martin Fränzle

Ernst-Rüdiger Olderog

Oliver Theel

Bernd Becker

Bernhard Nebel

Andreas Podelski

Christoph Scholl

Viorica Sofrie-Stokkermanns

Uwe Waldmann

Christoph Weidenbach

Reinhard Wilhelm

Verena Wolf

Three funding periods à 4 years

funding third period 2012-2015 9,6 Mill Euro

+ MPI

Page 9: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Ste

ering

Board

PA

Z

CVS Repository for Shared Sources

Benchmark Repository

Publication Repository

AVACS Web Site, Mailinglists

GForge Server for Cooperating Software Development

Beyo

nd

Tim

ed

Au

tom

ata

Tim

ing

An

aly

sis

, S

ch

ed

uli

ng

an

d

Dis

trib

uti

on

of

Real-

Tim

e T

asks

Heu

risti

c S

earc

h &

Ab

str

act

Mo

del

Ch

eckin

g f

or

Real

Tim

e T

asks

Co

nstr

ain

t B

ased

Veri

ficati

on

Fo

r H

yb

rid

Syste

ms

Au

tom

ate

d V

eri

ficati

on

of

Co

op

era

tin

g T

raff

ic A

gen

ts

Au

tom

ati

c V

eri

ficati

on

of

Hyb

rid

Syste

m S

tab

ilit

y

Co

mp

osit

ion

al A

pp

roach

es t

o

Syste

m V

eri

ficati

on

Dyn

am

ic C

om

mu

nic

ati

on

Syste

ms

Veri

ficati

on

of

Dep

en

dab

ilit

y

Pro

pert

ies

Pro

ject

Str

uct

ure

RA Coordinator

Fränzle

RA Hybrid

RA Coordinator

Podelski

RA Systems

Site Coordinator

Oldenburg WD

Site Coordinator

Freiburg BB

Site Coordinator

Saarbrücken RW

18 P

rincip

al In

vestigato

rs

RA Coordinator

Olderog

RA Real-Time

BB WD BF MF HH KM BN

ERO AP GP SR CS VS LT OT

UW CW RW

S1 B

F

S2 A

P

S3 H

H

R2 R

W

R3 B

N

R1 E

RO

H3 W

D

H4 O

T

H1/2

MF

Page 10: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Selected Highlights of Phase II

Page 11: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: System Structure

Reduce verification of parametrically generated systems to satisfiability of formula in decidable first-order theories

Demonstration on train application

Page 12: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: System Structure

Formal reduction of safety requirements in System of System application to requirements on local controllers

Demonstration Highway Entry Assistant

Page 13: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II:Branching/System structure

Formal model of cooperating transportation systems catering for failures, abstracted car dynamics, evolving shapes

Formal automatic synthesis of winning cooperation strategies

Demonstrated on Highway-Entry Assistance System

Page 14: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Branching Structure

Automatic computation of closed form solutions in parametric model-checking of stochastic systems

derive optimal parameters wrt time/reward/cost requirements fully symbolically

Page 15: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Branching Structure

Fully symbolic synthesis of winning strategies in timed-reachability games

Outperforms Uppaal-Tiga on standard benchmarks

Page 16: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Model Dynamics

Extending solvers for large boolean combinations of linear/non-linear/ transcendental functions

Ordinary Differential Equations

Stochastic constraints

Rich arithmetic

Page 17: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Model Dynamics

Decidability results

Quasi-decidability of hybrid system verification with non-linear dynamics

Parametric verification of an industrially relevant class of linear hybrid automata is in PTIME

Page 18: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Model Dynamics

Verification of timed systems with complex types

lists, arrays, pointers, sets √ uninterpreted functions over

primitive recursive functions reals satisfying monotonicity and boundedness properties

Page 19: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Specification Logics

Proof System for

Dynamic Hybrid Logic

Non-linear continous dynamics

Non-linear discrete updates

Evolving shapes

Demonstrated on Highway-Entry Assistance System

Page 20: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Specification Logics

Coordination logic

Logical representation for all decidable distributed realizability problems

Quantification over strategies with incomplete information

Explicates level of informedness given to strategies

Page 21: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Specification Logics

Specification Logic for SoS applications

First-order quantification over agents

Quantification over strategies with sets of agents

For time-bounded probabilistic reachability of SoS configurations

Demonstrated on Highway Entry Assistance Systems

Page 22: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Specification Logics

Reasoning about Remorse

Replacing the un-achievable concept of „winning strategy“ by new concept of remorse-free strategies: wrt a given world model and given set of observables, no other strategy can do better in comparable situations (i.e. environment moves)

Allows to define and test for optimal world models

Existence of remorse-free

strategies

Page 23: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Execution Platform Increased scope and precision of safe timing certificates

Distributed hierarchical inhomogenous bus architectures

Complex processors with out-of-order execution and speculation

Developed first formal notion of predictability and identified classes of predictable architectures

Formal approach for bridging the time gap

Anzahl der Steuergeräte: 75

Signalpfade: > 950

ICM-

Q/L

DME

Head-

unit

ZGW

NIVI

Infotainment

Komfort-Systeme

Zentrale

Steuerungssysteme

Assistenz und

Sicherheit

Antrieb

Fahrdynamik-Systeme

Page 24: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Scalability

Verification of timed automata with complex state spaces 300 fold improvement for coping with parallel composition

Fully symbolic and precise verification of hybrid systems with large discrete state spaces outperforming Phaver Dam controller with 11 real variables and 2100 discrete states verified

in 80 seconds

Tuning stochastic SMT solving for applications with up to 24 million discrete states and 23 real variables

Page 25: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Properties

System

Structure

Safety

Stability Bounded

response

Static

Existence of winning

strategies

Dynamic Highly

concurrent

Input

deterministic

Nondeterministic

Quantitative

Probabilistic

clocks Piecewise

Constant

rates

Differential

inclusions

Non-linear Linear Boolean

Discrete

Updates

Continuous

Evolution Bounded

arithmetic

Rich data Types

Virtual

Single

Processor

Distributed

homogenous

Distributed

Inhomogenous

Linear

arithmetic

Non-linear

arithmetic

Execution

Platform

Probabilistic

stability

Branching

Structure

Scalability

in #cont. var.

Scalability in

#discr. var.

Probabilistic

reachabiltiy

Alternating

Qualitative

Probabilistic

Liveness

CTL

Depth Bounded Safety

Duration

Selected Highlights Phase II: Scalability

Fully compositional approach for verification of safety and stability properties for hybrid controllers ( Transfer Project)

Heuristics for falsification of system requirements for timed automata yielding a three orders of magnitude improvement compared to previous phase

Page 26: AVACScmacs.cs.cmu.edu/presentations/AVACS_Survey.pdf · ALBERT - -LUDWIGS UNIVERSITÄT FREIBURG 2 Structure of Presentation The AVACS Vision Highlights of Phase II

- ALBERT-LUDWIGS

UNIVERSITÄT FREIBURG

Increasing Automation

• 67 tools supporting the AVACS approach to the analysis of complex systems

• see www.avacs.org/ tools

26


Recommended