A Layered Approach to Composition and Interoperation in Complex Systems Dr. Andreas Tolk Department of Engineering Management and Systems Engineering Old.

A Layered Approachto Composition and Interoperation

in Complex SystemsDr. Andreas Tolk

Department of Engineering Management and Systems EngineeringOld Dominion University, Norfolk, VA

Saikou Y. Diallo, Robert D. King, and Charles D. Turnitsa

Virginia Modeling Analysis and Simulation Center (VMASC)Old Dominion University, Norfolk, VA

Prepared for the HRA INCOSE Requirements Analysis and Management SeminarNovember 4, 2008, Newport News

Structure of the Presentation

• Levels of Interoperation– Currently applied Architecture Frameworks– Levels of Conceptual Interoperability Model

• Services, Agents, and Systems of Systems– What is needed to “understand” a component– What is needed to integrate a component

• Engineering Methods– Data Engineering– Process Engineering– Constraint Engineering

• Does it work?– Case Examples from NATO, Joint Forces Command, Department of Energy,

Department of Homeland SecurityHRA INCOSE Nov 2008, Newport News 2Tolk et al.: Layered Models

LEVELS OF INTEROPERATIONSection 1

HRA INCOSE Nov 2008, Newport News 3Tolk et al.: Layered Models

Framework Support• The task

– Integrate legacy solutions providing needed functionality in an aligned, orchestrated, and consistent way

• The answer– Service oriented architectures– Grid solutions– Federations of systems– System of systems

• The challenge– How to ensure aligned, orchestrated and consistent integration?– What frameworks and supporting artifacts are needed?


State of the Art: Current Focus

• Architectural Views– Functional– Physical– Operational

• System Views– Function– Structure– Behavior

• Unified Views (SysML, OPM, …)Focus lies still on Developing One System

HRA INCOSE Nov 2008, Newport News Tolk et al.: Layered Models 5

What is needed for SOSE?• Alignment

– Can data be obtained– Can data be mediated

• Orchestration– When to call which service/procedure– Integration of new functions into existing logical flow– Timing and synchronization– Post- and pre-conditions

• Consistency– Assumptions, constraints, and assertions


and all supported by Machines …

Levels of Conceptual Interoperability Model (LCIM)

Level 5 Dynamic Interoperability

Level 4Pragmatic Interoperability

Level 3Semantic Interoperability

Level 2Syntactic Interoperability

Level 0No Interoperability

Level 1Technical Interoperability

Level 6Conceptual Interoperability Increasing Capability for Interoperation

Modeling /Abstraction

Simulation /Implementation

Network /Connectivity


SERVICES, AGENTS, AND SYSTEMS OF SYSTEMS

Section 2


Technical Solutions• Service-oriented architectures

– Services are loosely coupled to provide needed functionality– XML, XSD, SOAP, UDDI, XSLT, …– OWL, OWL-S, …

• Agents– Intelligent Software Agents– Representing the functions/services– Collaborate with each other

• System of Systems


They should make life easier, but do they?

Artifacts needed

• In order to allow systems’ support, we need to make the functionality understandable for software– Web services needed– System components reusable– Agents representing functions correctly

• Challenge based on current support– Current solution focus on integration and interoperability

(technical challenges based on implementation)– Real interoperation requires conceptual alignment as well


Understanding in Models

The Three Premises for UnderstandingZeigler B.P. Toward a Simulation Methodology for Variable Structure Modeling, In Elzas/Oren/Zeigler (Eds.) Modeling and Simulation Methodology in the Artificial Intelligence Era, North Holland, 1986

PerceptionMeta-Models

RelationshipMapping

1 2

3

ObservedSystem

ObservingSystem



• The objective of ontologies is to document the conceptualization, which is another word for the result of the modeling process.

• This is done in a specified way, which means the application of engineering methods guided by rules and methods.

• The result is formalized, which means that machines and computers can not only read the result, but also make sense out of it in the context of their applications. Controlled

Vocabularies

Thesauri

Taxonomies

Ontologies

Logical Models

Applying the Ontological Spectrum

Semantic Web• Extensible Markup Language – XML

– XML enables data interchange between services and applications– XML supports the Syntactic Level of Interoperability by enabling a common structure of

data• Research Description Framework – RDF

RDF Schema – RDFS – RDF/RDFS enables data interchange between services and applications– RDF/RDFS supports the Syntactic Level of Interoperability by enabling a common

structure of data• Web Ontology Language – OWL

– OWL is based on XML and RDFS and supports therefore the Syntactic Level as they do– OWL was designed to support Strong Semantics supporting the Semantic Level in

machine readable form• OWL for Services – OWL-S

– The service defines the context of the data exchange, so that OWL-S supports Pragmatic Interoperability

– The Services Model of OWL-S can support Dynamic Interoperability, but the current versions (IOPE) do not deal with dynamic description of services in sufficient detail


Interoperability Contributions


Dyn

Pragm

Sem

Syn

none

Tech

Con

XML RDF (S) OWL OWL-S

ENGINEERING METHODSSection 3


Challenges of Interoperation

• Triangles of Interoperation– Scope– Resolution– Structure

– Conceptual model– Logical model– Physical model

– Data– Processes– Constraints


Can we make the system describing and organizing

the information exchange by itself?

Level 5 Dynamic Interoperability

Level 4Pragmatic Interoperability

Level 3Semantic Interoperability

Level 2Syntactic Interoperability

Level 1Technical Interoperability

Level 6Conceptual Interoperability

Cons

trai

nts

Data EngineeringProcess Engineering

Organizational and

Business Model


Data Engineering

• Data Administration– Data Administration identifies and manages the information exchange

needs between candidate systems (focusing on clearly defining the direction of data flow)

• Data Management– The goal of Data Management is to map concepts, data elements and

relationships from the source model to the target model.

• Data Alignment– The goal of data alignment is to identify gaps between the source and the

target.

• Data Transformation– The goal of Data Transformation is to align models in terms of their level of

resolution.


Process Engineering

• Process Cataloging– The important goal to achieve with cataloging is to gain an

understanding of where these processes are to be used

• Process Identification– Providing a description of what the process does, what its resource

and time requirements are to complete, and what data it operates on

• Process Alignment– Comparison of the information provided for two processes that are

part of the exchange of information for interoperability

• Process Transformation– Identify differences between processes and accommodate them by

middle-ware processesHRA INCOSE Nov 2008, Newport News Tolk et al.: Layered Models 19

Constraint Engineering

• Capturing assumption and constraints– The objective is to write down what the main concepts are

• Encoding Propositions– Encoded objectives in a knowledge representation language

• Comparing Assumption/Constraint Lists– Produce a measure of the semantic distance between propositions to

understand differences in machine coded form

• Adjudication and Resolution of Conflicts– Identify resolvable and irresolvable conflicts


CASES STUDIESSection 4


22Tolk et al.: Layered ModelsHRA INCOSE Nov 2008, Newport News

GE AdapterC2Linkopping

GE AdapterC2Linkopping

PABSTM&SMeppen

PABSTM&SMeppen

SICFC2Paris

SICFC2Paris

SIMBADM&SMadrid

SIMBADM&SMadrid

WebCOPC2Norfolk

WebCOPC2Norfolk

SitaWareC2Norfolk

SitaWareC2Norfolk

C-BML enabling Web Services

NATO MSG-027PATHFINDER Integration EnvironmentExperiment C2-M&S CouplingNovember 9, 2006


24SITAWAREHRA INCOSE Nov 2008, Newport News Tolk et al.: Layered Models

PABST SICF

SIMBAD

Joint Rapid Scenario Generation

Web Service

JEDIS DB

ASIT Client(TBB)

JIDPS Client(TBB)

UOBDAT Client(TBB)

SGS Init Client(TBB)

ACSIS

SGS

JIDPS

UOBDAT

JCATS Product Generator (JIDPS

TBB)

AWSIM Products Generator (SGS TBB)

JCATS Init Products

AWSIM Init ProductsJEDIS Web

Service

Authoritative Data Sources

Mediation Clients

Cohesive Data Product Repository

Data Consumers

Data Products

JEDIS Scenario Overview

JSAF Products Generator (SGS TBB)

JSAF Init Products


DOE and HLS

• In particular Data Engineering is recognized to be needed to gain a common understanding of operations– Several ontological works– Common vocabularies

• Idea of the LCIM applied in different contexts– GridWise Architecture Framework– Ontology tool development

• Presentations of LCIM and Data Engineering


SOME PROVOCATIVE IDEASFor the end


Do we still need Systems Engineering?

• System of Systems focus on cross-using functionality between legacy solutions

• Services encapsulate functionality for re-use in new contexts• System borders become more fluent• Requirements are valid until they are fixed, then they change immediately• New world: continuous flux of reuse and reconfiguration• Instead of Systems Engineering we need to educate for System of Systems

Engineering• System Engineering knowledge must be captured in machine

understandable form, as the lion share will be done by machines in the future

• System Engineers are needed, but they must to start to encode their knowledge


Literature• Andreas Tolk, Robert D. Aaron: “Data Engineering for Data-Rich Integration Projects:

Case Studies Addressing the Challenges of Knowledge Transfer,” Engineering Management Journal, in press

• Andreas Tolk, Charles D. Turnitsa, Saikou Y. Diallo: “Implied Ontological Representation within the Levels of Conceptual Interoperability Model,” International Journal of Intelligent Decision Technologies (IDT), Special Issue on Ontology Driven Interoperability for Agile Applications using Information Systems: Requirements and Applications for Agent Mediated Decision Support, Volume 2, Issue 1, pp. 3-19, January 2008

• Andreas Tolk, Saikou Diallo: “Model-Based Data Engineering for Web Services,” IEEE Internet Computing Volume 9 Number 4, pp. 65-70, July/August 2005

• Andreas Tolk, Saikou Y. Diallo, Robert D. King, Charles D. Turnitsa: “A Layered Approach to Composition and Interoperation in Complex Systems,” Chapter 3 in Tolk and Jain (Eds.): Complex Systems in Knowledge based Environments: Theory, Models and Applications. Series: Studies in Computational Intelligence, Vol. 168, Springer, 2009

• Andreas Tolk, Saikou Y. Diallo: “Model-based Data Engineering for Web Services,” Chapter 6 in Nayak et al. (Eds.): Evolution of the Web in Artificial Intelligence Environment, SCI 130, pp. 137–161, Springer, 2008


Questions and Comments

Dr. Andreas TolkAssociate ProfessorEngineering Management and Systems EngineeringOld Dominion UniversityNorfolk, VA 23529

Saikou Y. DialloRobert D. KingCharles D. TurnitsaVirginia Modeling Analysis and Simulation CenterOld Dominion UniversitySuffolk, VA 23435


A Layered Approach to Composition and Interoperation in Complex Systems Dr. Andreas Tolk Department of Engineering Management and Systems Engineering Old.

Documents

newport news slide

newport news tolk

system hra incose

system of systems hra

layered models10 slide

layered models5 slide

assertions hra incose

dynamic interoperability