Presentation is loading. Please wait.

Presentation is loading. Please wait.

Jeroen Voeten, Information and Communication Systems Group 1 System Level Modelling in POOSL J. Voeten, P. van der Putten,

Similar presentations


Presentation on theme: "Jeroen Voeten, Information and Communication Systems Group 1 System Level Modelling in POOSL J. Voeten, P. van der Putten,"— Presentation transcript:

1 Jeroen Voeten, email: voeten@ics.ele.tue.nl Information and Communication Systems Group 1 System Level Modelling in POOSL J. Voeten, P. van der Putten, M. Geilen, L. van Bokhoven and M. Stevens Information and Communication Systems Eindhoven University of Technology The Netherlands >

2 System Level Modelling in POOSL Information and Communication Systems Group 2 Contents Introduction: System Modelling The POOSL Language Formal Semantics Model Analysis Demonstration >

3 System Level Modelling in POOSL Information and Communication Systems Group 3 Introduction: System Modelling >

4 System Level Modelling in POOSL Information and Communication Systems Group 4 Design Decisions and Properties  Design involves the investigation of design alternatives to obtain a satisfying realisation.  Design decisions are based on the values of certain properties of alternative design realisations.  Realising design alternatives to determine the values of these properties is too costly and time-consuming.  Values of properties must be determined from models.  A property is a function on designs, realisations or models. If the range is Boolean the property is called qualitative (e.g. absence of deadlock), otherwise it is called quantitative (e.g. throughput). >

5 System Level Modelling in POOSL Information and Communication Systems Group 5 Models and Adequacy  A model is an abstract representation of some design realisation.  A model must be adequate (to a certain degree of accuracy) for the properties that are relevant for making a design decision.  A model is adequate (to a certain degree of accuracy) for a property if the difference between the property value of the model and that of the design realisation is within the accuracy bound. Vice versa, the design realisation is said to satisfy the model. >

6 System Level Modelling in POOSL Information and Communication Systems Group 6 Models and Abstraction  If a model is more abstract  it will in general be adequate for less properties with less accuracy;  it has a larger realisation space (collection of realisations that satisfy the model) and will be more difficult to realise;  it will be more compact;  it will take less time to develop it;  it will be cheaper to deduce property values.  Design decisions taken early in the design process (based on abstract models) have a large impact on property values. >

7 System Level Modelling in POOSL Information and Communication Systems Group 7 Towards System Models Increasing Expressive Power Abstraction Encapsulation > Less adequate and accurate More difficult to realise Design decisions have large impact More compact Less time to develop Cheaper to deduce property values HardwareSoftware TransistorsGatesRegisters, FSMsFunctions, DataObjectsAssembly CodeMachine CodeFunctions, Data Communicating (sub)systems System Behaviour RT Logic Transistor Object-Oriented HighAssembly Machine

8 System Level Modelling in POOSL Information and Communication Systems Group 8 Example: Switch Fabrics Decision Properties Throughput Average Delay Jitter Quality of Service Design Issues Queues Priorities Backpressure Scheduling Modelling Concepts Modelling Entities Concurrency Communication/Synchronisation Time/Stochasticity For different traffic types (CBR,VBR,ABR,UBR) >

9 System Level Modelling in POOSL Information and Communication Systems Group 9 Language Design Considerations/Alternatives Modelling Entities –Separate Data & Control  Autonomous Objects –Structural  Behavioural –... Concurrency –Synchronous  Asynchronous –Inherent to Modelling Entities  Orthogonality Communication/synchronisation –Synchronous  Asynchronous –Buffered  Unbuffered –... Time/stochasticity –Real-time  Discrete-time  Synchrony Hypothesis –Stochastic  Probabilistic  Non-determinism –... > Mathematical semantics Small Expressive Collection of Blending Language Primitives Selection Conflicts Orthogonality POOSL

10 System Level Modelling in POOSL Information and Communication Systems Group 10 The POOSL Language >

11 System Level Modelling in POOSL Information and Communication Systems Group 11 Overview of POOSL  POOSL (Parallel Object-Oriented Specification Language) is a formal specification language based on a timed version of process algebra CCS and on the basic concepts of traditional object-oriented programming languages (Smalltalk, Java, C++).  A POOSL specification consists of  A Top-Level Cluster;  Clusters & Cluster Classes;  Process Objects & Process Classes;  Data Objects & Data Classes. >

12 System Level Modelling in POOSL Information and Communication Systems Group 12 Example: Switch Fabric >

13 System Level Modelling in POOSL Information and Communication Systems Group 13 Clusters  Hierarchical structural entities;  Statically interconnected in a topology of channels;  Connect to the channels through private ports;  Consist of process objects and other clusters;  Behave asynchronous concurrent;  Communicate by synchronous message passing;  Organised in cluster classes. >

14 System Level Modelling in POOSL Information and Communication Systems Group 14 Example: Multistage Switch >

15 System Level Modelling in POOSL Information and Communication Systems Group 15 Process objects  Behavioural asynchronous concurrent entities;  Statically interconnected in a topology of channels;  Connect to the channels through private ports;  Communicate by synchronous message passing;  Contain data objects;  Compositional behaviour descriptions  Primitive statements: data,time,communication  Constructors: Parallel composition, interrupts, …  Behavioural abstraction: methods;  Are organised in process classes. >

16 System Level Modelling in POOSL Information and Communication Systems Group 16 Example: Switch >

17 System Level Modelling in POOSL Information and Communication Systems Group 17 Process Statements A method body consists of a process statement PS: The execution of PS can result inAction Performance & Time Passage PS ::= | ch!m(DE 1,…,DE n ) synchronous send | delay(DE) delay statement | while E do PS od repetition | DS data statement | if E then PS 1 else PS 2 fi selection message reception ch?m(p 1,…,p n | DE) | sel PS 1 or … or PS n les choice statement | par PS 1 and … and PS n rap parallel composition | PS 1 interrupt PS 2 interrupt statement | PS 1 abort PS 2 abort statement | [DE]PS guarded command | m(DE 1,…,DE n )(p 1,…,p n ) method call | PS 1 ;PS 2 sequential composition | timestamp x read model time

18 System Level Modelling in POOSL Information and Communication Systems Group 18 Example 1: Intuitive Semantics initialize()() ch!givePosition; delay(3.14); ch?position(point). ch initialize()() ch?givePosition; ch!position(point). instance variable point (10,12) instance variable point givePosition (10,12) position( ) copy > Time: 03.14

19 System Level Modelling in POOSL Information and Communication Systems Group 19 Example 2: initial method startUp of Switch startUp()() outputBufferArray:=new(Array) size(2); outputBufferArray put(1,new(BoundedFIFOBuffer) size(bufferSize)); outputBufferArray put(2,new(BoundedFIFOBuffer) size(bufferSize)); par handleInput1()() and handleInput2()() and handleOutput1()() and handleOutput2()() rap. >

20 System Level Modelling in POOSL Information and Communication Systems Group 20 Example 3: method handleInput1 of Switch handleInput1()() | cell:Cell | i1?cell(cell | outputBufferArray at(cell destinationPort) isNotFull) {outputBufferArray at(cell destinationPort) put(cell); cell nextDestinationPort}; delay(cellTime); handleInput1()(). >

21 System Level Modelling in POOSL Information and Communication Systems Group 21 Example 4: method handleOutput1 of Switch handleOutput1()() [outputBufferArray at(1) isNotEmpty] o1!cell(outputBufferArray at(1) firstElement) {outputBufferArray at(1) removeFirstElement}; delay(cellTime); handleOutput1()(). >

22 System Level Modelling in POOSL Information and Communication Systems Group 22 Example 5: Process Class Source >

23 System Level Modelling in POOSL Information and Communication Systems Group 23 Example 6: method of a TransmissionChannel transfer()() | p:Packet | in?packet(p); par delay(normal nextSample); out!packet(p) and transfer()() rap. Receive a new packet p from port in Wait for some stochastically determined amount of time before p is being delivered synchronously to port out Be simultaneously prepared to receive another packet >

24 System Level Modelling in POOSL Information and Communication Systems Group 24 Example 7: timeouts and watchdogs waitForInput()(value,status:String) ch!inputRequest; sel ch?input(value);status:=‘ok’ or delay 20;status:=‘timeout’ les Wait for input on port ch. Return from the method call if an input is not received within 20 time units > timeCriticalComputation()() doComputation()() abort (delay deadline; errorHandling()()). Call method doComputation()(). If the method has not terminated within a deadline amount of time, abort it and start an errorHandling routine.

25 System Level Modelling in POOSL Information and Communication Systems Group 25 Data objects  Behavioural sequential entities;  Can be created dynamically;  Communicate by message passing;  Invoke methods upon message reception;  Return results of method invocations to sender;  Are organised in data classes. >

26 System Level Modelling in POOSL Information and Communication Systems Group 26 Example: Data class Exponential >

27 System Level Modelling in POOSL Information and Communication Systems Group 27 > Data Statements and Expressions DS ::= x:=DE | DS 1 ;DS 2 | while DE do DS | if E then DS 1 else DS 2 fi | DE assignment to variable or parameter sequential composition repetition selection data expression A method body consist of a data statement DS : > DE ::= x | new(C) | self | DE m (DE 1,…,DE n ) | -1,0,’a’,’b’,3.14,true,false | nil data object referenced by x newly created data object of data class C data object evaluating this expression method call constants of primitive classes constants undefined data object DE is a data expression, always evaluating to a data object:

28 System Level Modelling in POOSL Information and Communication Systems Group 28 mu rand withParam(m:Real):Exponential mu:=m; rand:=new(RandomGenerator); return(self). Example: Intuitive Semantics Consider the execution of the following statement: exp:=new(Exponential) withParam(3) 3 some Random >

29 System Level Modelling in POOSL Information and Communication Systems Group 29 Formal Semantics >

30 System Level Modelling in POOSL Information and Communication Systems Group 30 Formal Semantics: Transition System Plotkin-style Structural Operational Interleaving Semantics defines a probabilistic labelled transition system: Model M can perform action a with probability p and then behave as M’ Model M can delay for time t with probability p and then behave as M’ > (Mod,Act,T,{ | a  Act,p  [0,1]},{ | t  T,p  [0,1] }) a,pt,p M M’ a,p Compositional Definition with Axioms and Inference Rules M M’ t,p delay(t)  t,1 M 2 M’ 2 t,p 2 M 1 M’ 1 t,p 1 M 1 || M 2 M’ 1 || M’ 2 t,p 1 p 2

31 System Level Modelling in POOSL Information and Communication Systems Group 31 Action Urgency: Two-Phase Model [ X.Nicollin, J.Sifakis ’91] Asynchronous Execution of Actions Synchronous Passage of Time >

32 System Level Modelling in POOSL Information and Communication Systems Group 32 Example: Lossy Channel >

33 System Level Modelling in POOSL Information and Communication Systems Group 33 Transition System of a Lossy Channel >

34 System Level Modelling in POOSL Information and Communication Systems Group 34 Model Analysis >

35 System Level Modelling in POOSL Information and Communication Systems Group 35 Analysis Validation – Are we building the correct model Qualitative Verification – Analyse correctness properties Quantitative Verification – Analyse performance properties Synthesis – for rapid model analysis – for implementation/realisation >

36 System Level Modelling in POOSL Information and Communication Systems Group 36 Validation >

37 System Level Modelling in POOSL Information and Communication Systems Group 37 Qualitative Verification: Equivalence Checking Abstraction to CCS Minimisation Equivalent minimised protocol >

38 System Level Modelling in POOSL Information and Communication Systems Group 38 Qualitative Verification: Model Checking Specify the property (  ) to verify in a temporal logic (LTL). Example: A stimulus S always results in a response R within 10 ms Construct an automaton that accepts precisely the traces satisfying this property. Check whether the model satisfies the property –Exhaustively Generate complete transition system (T) of the model Check for language inclusion L(T)  L(  ) or for empty intersection L(T)  L(  ) =  Proof of correctness but only applicable in case of relatively small finite-state systems. –Non-exhaustively Generate one execution trace from transition system (T) Check whether some infinite completion of this trace can be accepted by the automaton No proof of correctness, but always applicable to detect errors. >

39 System Level Modelling in POOSL Information and Communication Systems Group 39 Quantitative Verification: Performance Analysis Specify the performance metric as a (temporal) reward. Examples: jitter, buffer fill levels. Determine the long-run average metric value –Analytically Generate the complete probabilistic labelled transition system Interpret it as a Markov chain with (temporal) reward structure Compute performance metric by using the strong law of large numbers and the ergodic theorem for Markov chains Certain/precise results but only applicable in case of relatively small finite- state systems. –Empirically Generate one execution trace Estimate the metric value, confidences and errors using the strong law of large numbers and the central limit theorem for Markov chains Uncertain results but applicable in case of large and even infinite-state systems. >

40 System Level Modelling in POOSL Information and Communication Systems Group 40 Synthesis To analyse large systems, it is important to be able to generate the model’s transition graph or a trace thereof rapidly. Automatic mapping from POOSL to C++ using Process Execution Trees: –Self-modifying data-structure representing the state of the model; –Compositional mapping of all language constructs; –Offering facilities for model debugging; –No use of (real-time) operating system; –Slave-like C++ module facilitating the interfacing with other tools; –Automatic storage reclamation. >

41 System Level Modelling in POOSL Information and Communication Systems Group 41 Process Execution Trees root ; sel granted c?mc!m finishedkill finished root ; sel c?mc!m startUp granted startUp State n n+1 State n+2 talk()() sel c!m or c?m les; talk()()

42 System Level Modelling in POOSL Information and Communication Systems Group 42 Demonstration >

43 System Level Modelling in POOSL Information and Communication Systems Group 43 The Validation Tool SHESim Tool Demonstration Example of a simple datalink protocol Download tool via ftp at: ftp.ics.ele.tue.nl/pub/projects/she_sim/prac9900 >


Download ppt "Jeroen Voeten, Information and Communication Systems Group 1 System Level Modelling in POOSL J. Voeten, P. van der Putten,"

Similar presentations


Ads by Google