Chương 13: Modeling System Behaviours doc

www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons
Building System Models for RE
Building System Models for RE
Chapter 13
Modeling System Behaviours
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 2
Building models for RE
Behaviors -
Behaviors -
Scenarios
Scenarios (Chap.13)
Behaviors -
Behaviors -
State machines
State machines (Chap.13)


www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 3
The behaviour model

Behaviour view of the system being modeled
–
Modelling the dynamic of an agent, consider:
Modelling the dynamic of an agent, consider:
Specific behaviours of specific agent instances
All possible behaviours of any agent instance.

Specific instance behaviours are captured through scenarios.

General class behaviours are captured through state machines.
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 4

Modeling system behaviours: outline

Modeling instance behaviours
– Scenarios as UML sequence diagrams
– Scenario refinement

Modeling class behaviours
–
State machines as UML state diagrams
–
State machine refinement

Building behaviour models
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 5
Modeling instance behaviours

What is scenario?
What is scenario?
–
Is a temporal sequence of interaction events among agent instances
–
Positive scenario:
Positive scenario: the sequence of interactions illustrates a possible
way of satisfying an behavioural goal.
–
Negative scenario:
Negative scenario: the sequence of interactions illustrates a possible
way of satisfying an obstacle to a goal.

Interaction

Interaction:
– Is an instantaneous object
–
Corresponding to the source agent instance applying an
operation whose effect is monitored by the target agent
instance.
–
Source agent controls the interaction; target agent monitors it.
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 6
Scenario as UML sequence diagrams

Sequence diagrams are a UML variant of the event trace diag.

Basic UML syntax for sequence diagrams is similar to the one used in
event trace diagrams.
Figure 13.1 – Scenario represented by a sequence diagram

: Train
Actuator/Sensor

: OnBoard
Controller

doorsOpening

exit

doorsClosing

p1: Passenger

entrance


p2: Passenger

software agent



instance



environment agent



time



interaction



event name

! trainStopped

startAcceler

www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 7
Scenario as UML sequence diagrams (2)
Figure 13.2 – Sequence diagram for a loan scenario

: LoanManager
BookRequest (PatrID, BookID)

: Staff
self-interaction



event attributes



instance disappears




: CopyManager
LoanQtyOK? (PatrID)

CpyAvailable? (BookID)

Reserved? (BookID)

OK-Available (CopyID)

OK-Book (PatrID, CopyID)

checkOut (PatrID, CopyID)

Registered? (PatrID)

Figure 13.3 – Negative scenarios and optional interactions

opt

: Train
Actuator/Sensor
: OnBoard
Controller
AlarmPressing

AccelerComnd

doorsOpening
prohibited
interaction



neg

: Train
Actuator/Sensor
doorsOpening
exit

doorsClosing

:Passenger
optional
interaction



: Passenger

: OnBoard
Controller
! trainStopped
startAcceler

AlarmPropagation

www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 8
Scenario refinement: Episodes and agent
decomposition

When elaborating complex scenarios stepwise, expressing a
When elaborating complex scenarios stepwise, expressing a
coarse-grained scenario first then refine it with further details

coarse-grained scenario first then refine it with further details

coarse-grained scenario -> finer-grained scenario:
coarse-grained scenario -> finer-grained scenario:
–
Introducing episodes
Introducing episodes
–
Decomposing an agent instance into finer-grained agents.
Decomposing an agent instance into finer-grained agents.
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 9
Scenario refinement: Episodes
Figure 13.4 – Scenario refinement: introducing episodes

:Scheduler
meetingRequest
(dateRange, withWhom)

: Initiator
reference to
another
diagram




: Participant
schedule
Determination

ConstraintsAcquired

notification
(date, location)

notification
(date, location)

:Scheduler
? constraints
(dateRange)


: Participant
OK-constr

! constraints

sd ConstraintsAcquired

episode



OK-request


www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 10
Scenario refinement: agent decomposition
Figure 13.5 – Scenario refinement: agent decomposition

: LibraryManager
BookRequest
(PatrID, BookID)

: Staff
RequestOK?
(PatrID, BookID)

OK-Book
(PatrID, CopyID)

: LoanManager
: Staff

: CopyManager
CpyAvailable? (BookID)

Reserved?
(BookID)

OK-Available (CopyID)

checkOut (PatrID,CopyID)

BookRequest
(PatrID, BookID)

OK-Book
(PatrIDCopyID)

LoanQtyOK? (PatrID)

Registered? (PatrID)

www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 11
Modeling system behaviours: outline

Modeling instance behaviours
– Scenarios as UML sequence diagrams
– Scenario refinement

Modeling class behaviours
–
State machines as UML state diagrams
–
State machine refinement

Building behaviour models
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 12
Modeling class behaviours

State machines complement the fragmentary information
provided by scenarios in multiple ways:
– They make state information explicit
–
They capture the behaviour of any agent instance, not just a
specific one.
–
They are aimed at capturing all admissible sequences of
state transitions, not just a specific one.


Snapshot state <-> SM state

State transitions captured by a state machine refer to SM states.

The events causing state transitions can be of different types:
–
External events: the agent associated with the State Machine
does not controls.
–
Internal events: events controlled by the agent associated with
the State Machine
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 13
State machines as UML state diagrams

A state machine is represented in UML by a variant of a statechart
called a state diagram.

Some features …
–
State transition is caused by an event
–
Transition may be labeled by the name of the causing event
– In contrast to events, a state has a duration.
–
Initial and final states
Figure 13.6 – A simple SM diagram for a BookCopyInfo entity controlled by LibraryManager

Available

onLoan


checkOut

(PatrID, self)


return

(PatrID, self)


event with
attribute


loss


SM state



initial state



final state

transition



www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 14
State machines as UML state diagrams (2)

Guarded transitions
–
A guard captures a necessary condition for transition firing

Actions
–
An action is an auxiliary associated with a state transition.

Event notifications
–
An important class of action where an event is notified by a
‘producing’ diagram to a ‘consuming’ diagram where this
event causes transitions.
– This allows for synchronization among multiple diagrams.

Entry/exit actions
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 15
State machines as UML state diagrams (3)
Figure 13.7 – A SM diagram for a BookInfo entity controlled by LibraryManager

Borrowable

UnBorrowable

copyBorrowing
[not LastCopy ]

copyReturn

copyBorrowing
[ LastCopy ]

guard



copyReturn

Figure 13.8 – Auxiliary actions in a state diagram

doorsClosed

doorsOpen


doorsOpening


doorsClosing


[Speed = 0 and AtPlatform]

/ Display terminalInfo


/ Activate warningRinging
action


www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 16
State machines as UML state diagrams (4)
Figure 13.9 – Send actions for diagram synchronization: BookCopyInfo revisited

Available

onLoan


checkOut

(PatrID, self)


return

(PatrID, self)


event
notification

to consumer
loss


/ send BookInfo.copyReturn


/ send BookInfo.copyBorrowing


Figure 13.10 – Avoiding duplicated actions: entry and exit actions within a state

OK-request
/ Inform initiator


…


ConstraintsRequested
entry / Inform initiator


Planning


Resolving


[all available]

[conflicts]


weakeningRequest

/ Inform initiator
ValidatingMeetingData
…


entry
action


www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 17
State machine refinement: sequential and concurrent
sub-states

… when we elaborate a complex state machine model stepwise:
– specify first a coarse-grained state diagram
–
refine this diagram with more states that are finer-grained
– Sub-state -> nested state / super-state -> composite state

Sequential decomposition
–
An SM state may be decomposed into sub-states
– Structuring mechanism is recursive

–
Sub-state my in turn be sequentially decomposed

Parallel decomposition
–
An SM state may be decomposed into concurrent sub-states
–
Composite state can be structured as a parallel composition
of nested diagrams.
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 18
State machine refinement: sequential decomposition
Figure 13.11 – Sequential state decomposition for the TrainInfo entity controlled by TrainController

Stopped

Moving
include / movingSubstates


[Speed = 0]


Accelerating

Decelerating


(a)

AccelerComnd


[Acceler
≤
0]


movingSubstates


[Acceler
>
0]


AccelerCommand


AccelerComnd


[Acceler
>
0]


AccelerComnd

[Acceler
≤
0]


startAcceler


[Acceler
>
0]


Stopped

[Speed = 0]


startAcceler


[Acceler
>
0]


AccelerComnd


[Acceler

>
0]


Moving
Accelerating

Decelerating


AccelerComnd


[Acceler
≤
0]


AccelerComnd


[Acceler
≤
0]


AccelerComnd


[Acceler

>
0]


(b)


reference


sequential substate

initial substate


Semantic rules define sequential state decomposition more precisely:
[textbook, p459)
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 19
State machine refinement: parallel decomposition

Semantic rules define parallel state decomposition more precisely:
[textbook, p461)
Figure 13.12 – Parallel state decomposition and hierarchical state machines

TrainInfoState


DoorsState

MovementState


parallel
composition


state name


composite state


Stopped

[Speed = 0]


startAcceler


[doorsState = closed]


AccelerComnd


[Acceler
>
0]

Moving
Accelerating

Decelerating


AccelerComnd


[Acceler
≤
0]


AccelerComnd


[Acceler
≤
0]


AccelerComnd


[Acceler
>
0]

doorsClosed

doorsOpen


[Speed = 0 and AtPlatform]


doorsOpening


/ Display terminalInfo


doorsClosing


/ Activate warningRinging
TrainInfoState


DoorsState


MovementState


sequential

substate


concurrent
substate

guard as
synchronizing
condition


+ sequential
decomposition

www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 20
Modeling system behaviours: outline

Modeling instance behaviours
– Scenarios as UML sequence diagrams
– Scenario refinement

Modeling class behaviours
–
State machines as UML state diagrams
–
State machine refinement

Building behaviour models
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 21
Building behaviour models


Goals, scenarios and state machines have complementary
strengths and limitations:
– Goals are declarative, capture functional, non-functional and
alternative options aspects.
–
Scenarios support an informal, narrative and concrete style
of expression.
–
State machines provide visual abstractions of explicit
behaviours of any agent instance in a class.

Expecting goals, scenarios and state machines to form a win-win
partnership for building complex models.

Table 13.1 summarizes the respective pros/cons of goals,
scenarios and state machines. (textbook.p464).
www.wileyeurope .com/college/van lamsweerde Chap.13: Modeling System Behaviours © 2009 John Wiley and Sons 22
Building behaviour models

Elaborating relevant scenarios for good coverage

Decorating scenarios with state conditions

From scenarios to state machines

From scenarios to goals

From operationalized goals to state machines.