18
Introduction Tools Towards Print Production Simulation Giordano B. Beretta Print Production Automation Lab Hewlett-Packard Laboratories Palo Alto, California 27 May 2010 G. Beretta PPAL Commercial Print Automation Project

Towards Print Production Simulation

Embed Size (px)

DESCRIPTION

Basic concepts required for working on print production simulation

Citation preview

Page 1: Towards Print Production Simulation

IntroductionTools

Towards Print Production Simulation

Giordano B. Beretta

Print Production Automation LabHewlett-Packard Laboratories

Palo Alto, California

27 May 2010

G. Beretta PPAL Commercial Print Automation Project

Page 2: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Conventional Printing

Printing is an industrial manufacturing processKey to success is an optimized workflowWorkflows have been been perfected over centuriesEmploy highly skilled workersInterface: purchasers and sales representatives

purchasing

fulfillment plant managerpress operator

graphic artist

sales rep

art directorclient

contract

G. Beretta PPAL Commercial Print Automation Project

Page 3: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Print Services

Paradigm shift enabled by digital printingCombination with e-commerce and services (cloudcomputing); run length can be as low as 1Instead of saving work, labor-saving devices permiteverybody to do their own work (McLuhan)Artwork creators are no longer highly skilled

print service provider

pupil

office worker

consumerservice license

agreement

print service

aggregator

G. Beretta PPAL Commercial Print Automation Project

Page 4: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Conventional vs. PSP

metric conventional PSP

creator designer laymen + web servicecustomer direct indirect contactrun length > 500 > 1scheduling simple complexmotivation business vanity, hobbymarketing CRM social networkexpectations realistic highchallenge cost manage expectations

G. Beretta PPAL Commercial Print Automation Project

Page 5: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Challenges

Known:Workflow optimizationMass customizationFinishing

New:Complex heterogeneous systemEmergent model behavior

behavior caused by the interactionbetween characteristics of differentmodels and not intended or foreseenby the modelexample: priority inversion betweenthreads in a real-time OSleads to the invention of newscheduling mechanisms

G. Beretta PPAL Commercial Print Automation Project

Page 6: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Main Point

To a print job a print plant looks like a pipelineTo a print plant a stream of print jobs is a collection ofprocessing steps that must be performed in parallel whileavoiding idle times on resources schedulingThe key challenge is to study emergent behaviorEmergent behavior is discovered through simulation — theexecution of models

Definition (amorphous modeling)

The style of handling heterogeneous models that allowsvarious interaction mechanisms to be specified between agroup of components at the same time is called amorphous

G. Beretta PPAL Commercial Print Automation Project

Page 7: Towards Print Production Simulation

IntroductionTools

A Paradigm Shift in PrintingMain PointAgenda

Agenda

Build formal amorphous models ofprint service providers (PSP)Discover emergent behavior throughsimulationStudy emergent behavior throughtrace and debug of real PSP dataStudy PSP data through data miningand classification using machinelearning

G. Beretta PPAL Commercial Print Automation Project

Page 8: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Typesetter Trade Union

Kristen Nygaard (1957) and Ole-Johan Dahl (1962)First (1962) there was Simula I, a superset of Algol 60, forsimulating typesetting (discrete event networks)Then there was Simula 67 with objects, classes,subclasses, virtual methods, co-routines, discrete eventsimulation, and garbage collectionFrom Oslo to Palo Alto: Smalltalk, Interlisp, Cedar,Modula-2, Oberon — adding threads (true concurrency,not just co-routines)Side-trip to C++ and Standard Template LibraryFinally: Java and C#

G. Beretta PPAL Commercial Print Automation Project

Page 9: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Actor-Oriented Design

Hewitt & co-worker’s 1973 extension to exploit massivelyparallel computersConcurrent execution of objectsCommunication by message passing

DefinitionAn actor is a computational entity that, in response to amessage it receives, can concurrently:

send a finite number of messages to other actorscreate a finite number of new actorsdesignate the behavior to be used for the next message itreceives

Communication is decoupled from senderG. Beretta PPAL Commercial Print Automation Project

Page 10: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

What is Different?

First there were interrupt handlers and co-routinesBasic idea of threads: shared memory and global state

semaphores (Dijkstra), monitors (Hoare), serializer (Hewitt& Atkinson)

In the actor model, messages are simply sentno buffering, no synchronous handshakingno ordering, FIFO requires explicit queue actoreverything is localinfluenced by packet switched networks, PUP

Message can contain another actor (e.g., resumption,a.k.a. continuation, stack frame) to which recipient sendsresponse variable topologyBehavior: mathematical function to express what an actordoes when it processes a message

G. Beretta PPAL Commercial Print Automation Project

Page 11: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Ptolemy II — What is it?

A laboratory for experimenting withdesign techniques of concurrentembedded systems that mixtechnologies, for exampleelectronics and mechanicaldevices

G. Beretta PPAL Commercial Print Automation Project

Page 12: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Modeling

DefinitionModeling is the act of representing a system or subsystemformally

mathematical a set of assertions about properties of thesystem such as its functionality or physicaldimensions

constructive defines a computational procedure that mimics aset of properties of the system; also calledexecutable model or simulation

G. Beretta PPAL Commercial Print Automation Project

Page 13: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Design

DefinitionDesign is the act of defining a system or subsystem

Usually this involves defining one or more models of the systemand refining the models until the desired functionality isobtained within a set of constraints

G. Beretta PPAL Commercial Print Automation Project

Page 14: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Embedded Software

DefinitionEmbedded software is software that resides in devices that arenot first-and-foremost computers

A key feature of embedded software is that it engages thephysical world, and hence has temporal constraints thatdesktop software does not share

G. Beretta PPAL Commercial Print Automation Project

Page 15: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

MoC and Semantics

Definition (MoC)Executable models are constructed under a model ofcomputation, which is the set of “laws of physics” that governthe interaction of components in the model

Definition (Semantics)A set of rules that govern the interaction of components iscalled the semantics of the model of computation

G. Beretta PPAL Commercial Print Automation Project

Page 16: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Programming Paradigm

Most (but not all) Ptolemy II MoCs support actor-orienteddesign (see slide 9)The Ptolemy II abstract syntax is described by models,actors, ports, parameters, and channelsA Ptolemy II model is a set of actors, ports, attributes, andconnectionsProgramming is done graphically, via XML documents, orin a conventional programming languagePtolemy II features:

large family of models of concurrencynot all actors are threadscommunication needs not be strictly asynchronousmodularity (inner classes, multiple inheritance)

G. Beretta PPAL Commercial Print Automation Project

Page 17: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Data Mining

DefinitionData mining is the process of extracting patterns from data

Difference between data mining and statistics:Data mining uses machine learning to analyze patterns inlarge data sets, typically stored in databases

data mining aims at uncovering hidden patternsStatistics refers to the collection, analysis, interpretation orexplanation, and presentation of smaller data sets, typicallystored in arrays

statistics aims at testing hypotheses

G. Beretta PPAL Commercial Print Automation Project

Page 18: Towards Print Production Simulation

IntroductionTools

Basics ConceptsPtolemy II ConceptsData Mining Concepts

Data Mining Tasks

Data mining commonly involves four classes of tasks:Clustering is the task of discovering groups and structures in

the data that are in some way or another “similar,”without using known structures in the data

Classification is the task of generalizing known structure toapply to new data (common algorithms includedecision tree learning, nearest neighbor, naiveBayesian classification and neural networks)

Regression Attempts to find a function which models the datawith the least error

Association rule learning Searches for relationships betweenvariables

G. Beretta PPAL Commercial Print Automation Project