modelling reactive systems 4 professor muffy calder dept. of computing science university of glasgow...

11
Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow [email protected]

Upload: kathleen-harrell

Post on 18-Jan-2016

220 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Modelling Reactive Systems 4

Professor Muffy CalderDept. of Computing Science

University of Glasgow

[email protected]

Page 2: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Aims

To introduce and explore 2 different formalisms and analysis techniques used in the design of reactive systems.

To explore the practical issues raised by using these formalisms and associated software tools, particularly within the context of communications systems and protocols.

Includes:

state-based modelling using Promela language

reasoning using model-checker SPIN

linear temporal logic

event-based modelling using LOTOS

communication mechanisms

lots of concurrency - in practice and some theory

Exercise:

model telephone system

reason about system using SPIN

Page 3: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

References

My notes (on web)

Technical papers

Web pages (for both LOTOS and Promela)

Two (out of print!) books

Page 4: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Modelling Reactive Systems

Reactive system

• event driven, internal or external stimuli

• real-time, concurrent, or distributed

Examples

• telephony

• communications networks

• operating systems

• avionics

• VLSI

• HCI

• railway signalling

• others (from biochemistry to cat flaps!)

Page 5: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Modelling Reactive Systems

Features

• nondeterminism

• concurrency

• communication

Concerns

• fairness

• race-conditions

• timing

• safety properties

(something bad does not happen)

• liveness properties (something good happens)

(Liveness is harder to prove than safety)

Page 6: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

State or Event-Based Modelling?

What is a process?

How do processes interact?

• is communication synchronous

• is communication asynchronous

How do we model a process?

• event based

• state-based

Page 7: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Event based a

b c

stop stop

State-based

ev== a

e== b

ev== c ev= =d

Events label edges

Variables label states

a; ((b; stop) [] (c; stop))

ev = a;ev = b;if :: ev= c :: ev = dfi

LOTOS

Promela

Page 8: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Finite State Automata (informal)

FSA is conceptual machine

• a given number of states (finite or infinite)

• m/c is in exactly one state, at any time

• state transitions are caused by (input) events

• state transition may cause (output) events

• typically, states represent visible modes of behaviour.

STS structured transition system

• graphical representation of an FSA

• nodes represent states

• edges represent events

Examples: a chocolate machine

Page 9: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Kripke Structures (informal)

A Kripke structure consists of

• a given number of states

• an initial state

• a set of final/accepting states

• a transition relation between states

• a function labelling states with valuations of

the variables in states

A computation path is a sequence of states s0,s1,… such that there is a transition between si and si+1.

A computation path may be infinite.

We are concerned with systems which have a finite number of states, but computation paths may be finite or infinite.

Page 10: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Example

• A chocolate dispensing machine.

Page 11: Modelling Reactive Systems 4 Professor Muffy Calder Dept. of Computing Science University of Glasgow muffy@dcs.gla.ac.uk

Process Diagrams

A channel is a directed communication link between 2 processes.

A message or signal may by sent and received to/from a channel.

A process diagram:

A static description of system/network “topology”.

Do not confuse these with automata!

Process a

Process b

message

message

message

Environment