High Level Architecture COTS Simulation Package Interoperability Forum HLACSPIF

1
High Level Architecture-COTS Simulation Package
Interoperability Forum (HLA-CSPIF)

• Simon Taylor
• Centre for Applied Simulation Modelling
• Department of Information and Computing
• Brunel University, London, UK
• simon.taylor_at_brunel.ac.uk
• Supported by the UK EPSRC GROUPSIM Network
• (COTS Commercial-off-the-shelf)

2
Acknowledgements

• Mark Elder Visual Thinking International
  (Simul8)Gan Boon Ping SIMTech,
  Singapore.Nirupam Julka SIMTech, SingaporeJohn
  Ladbrook Ford Motor Company.Peter
  Lendermann SIMTech, Singapore.Malcolm
  Low SIMTech, Singapore.Stewart Robinson Warwick
  Business School, UK.Rick Severinghaus SISOSteffe
  n Strassbruger DaimlerChrysler, GermanySteve
  Turner Nanyang Technological University,
  SingaporeAlexander Verbraeck, Delft
  University of Technology, NetherlandsTony
  Waller Lanner (Witness)

3
Briefing

• Background Motivation
• COTS Simulation Modelling (Modeling Simulation)
• Examples
• HLA-CSPIF
• Reference models and compliancy
• types
• not dealt with real time information/ historical
  information (yet)
• Conclusions

4
Background and Motivation

• Co-chair UK ORS Simulation Study Group (1996)
• Workshops in various aspects of simulation
  modelling
• gtGROUPSIM Network (2000)
• Developing strategies and infrastructures for
  collaborative simulation modelling
• gtHLA-CSPIF (2002)
• Standardization of COTS simulation package
  interoperability
• gtHPM-net (2003)
• Human performance modelling network

5
Background and Motivation

• Low take up of distributed simulation in
  non-defense domains (manufacturing, logistics,
  business, supply chains, etc.)
• Low demand BUT growing rapidly (lack of
  non-technological examples, technological
  counselling)
• Much excellent work on-going to provide
  distributed simulation solutions but little
  common ground between technologists,
  simulationists and package vendors
• gt HLA-CSPIF

6
COTS Simulation Modelling

• Marker to encapsulate the principles and
  practices of those who use and supply visual
  interactive modelling environments and languages
• Analysis, training, facilitation
• COTS packages (about 30 available)
• Cost from 400 to 15,000
• Common set of modelling elements variants
• Vary in functionality (Animation, VR,
  Optimisation, etc.)

7
(No Transcript)
8
(No Transcript)
9
(No Transcript)
10
Generic Models
Push
Pull
How do we interoperate models hosted by COTS SPs?
11
(No Transcript)
12
(No Transcript)
13
How do we interoperate models hosted by COTS SPs?
Bing!
STANDARDS
14
HLA-CSPIF
(25 members,7 countries)(SISO Reflector)

• The aim of the HLA-CSPIF is to create a
  standardized approach to distributed simulation
  using the IEEE 1516-2000 High Level Architecture
  to support the interoperation of discrete event
  models created in COTS simulation packages in a
  manner that is transparent to the end user
• The objectives of HLA-CSPIF are to
• Involve representative researchers,
  simulationists and vendors in the standardization
  process,
• Create artefacts that can be used to communicate
  concepts and problems between researchers, users,
  and vendors in support of the HLA-CSPIF aim,
• Develop a standard data exchange representation
  (OMT),
• Develop a standard data exchange mechanism
  (RTI/Federate Ambassadors),
• Develop a standard specification for a
  distributed simulation co-ordination/experimentati
  on tool (federate), and
• To publicise the forum

15
Progress

• Investigation of data exchange requirements (OMT)
  (cf RPR FOM)
• Five workshops over past year
• COTS simulation models can potentially exchange a
  wide range of information which present a wide
  range of technical challenges
• Reference models are being developed to reduce
  the complexity of the solution space
• Solutions can therefore be classified on the
  basis of successful compliance

16
Reference Models

• Derived predominantly from discussion with
  simulationists and package vendors (i.e. they
  address real problems)
• A reference model is intended to be the simplest
  representation of a problem to promote a common
  solution set and CLEAR understanding of the
  problem to simulationists, package vendors and
  technologists
• Type I Entity passing
• Type II Entity passing with bounded buffer
• Type III Global resources
• Type IV Global events
• Type V Shared conveyors

17
Basic Model

• Indicative of many different process modelling
  problems (Qn,Wn)
• Not all packages have sources and sinks but do
  have the notion of entry and exit

18
Type I Entity passing

• Timestamped entities arrive at So2
• Sources and sinks are transparent
• Local/distributed implementation should be
  transparent to the end user

19
Type II Entity passing with bounded buffer

• Q2 is bounded. W1 must block when Q2 is full.
  When Q2 has space, W1 must unblock.

20
Reference FOM BAL (Type I/II)

• Spreadsheet people gt feedback dependency
• Unbounded/Bounded behaviour
• Lookahead issues

21
Type III Global Resource

• Resource R is shared between the two models as a
  global resource
• A change in R1 must be synchronised with R2 and
  vice versa

22
Type IV Global Event

• A global event E occurs in all models at the same
  simulation time
• The existence of E1 implies E2 and vice versa

23
Type V Shared Conveyor

• Conveyor C is shared/split between the two models
  so that the exact position of each entity is
  shown correctly

24
Current focus -gt OMT Issues (Type I, II)

• Basic problem is although entity representation
  is trivial, conflicting representations are
  incompatible
• gtneed a standard approach
• Entities consist of
• name (compatibility)
• set of attributes with datatypes (which may not
  be supported by all COTS packages and can be
  complex)
• semiotic values (documentation?)
• timestamp (time resolution may vary)
• Representation options
• attributes?
• attributes/complex data type?
• interaction/parameter complex data type?

25
Conclusions

• Type I models will be dealt with in August 2003
  and if possible put forward as a candidate for
  SISO productization activity approval
• Fall SIW 2003 Update
• DS-RT 2003/ESS 2003 23rd-29th October 2003
  (Delft, The Netherlands)
• HLA-CSPIF/XMSF Workshop
• Papers can still be submitted
• WSC 2003
• Distributed simulation mini-track panel
• UK Simulation Workshop 2004
• Much still to do!
• Keep in touch through the SISO reflector!

26
Annex
27
OMT Tables

• Object Class Structure Table.
• Records the namespace of all federate or
  federation object classes.
• Interaction Class Structure Table
• Records the namespaces of all federate or
  federation interaction classes.
• Attribute Table
• Specifies features of object attributes in a
  federate or federation.
• Parameter Table
• Specifies features of interaction classes
  parameters in a federate or federation.
• Datatype Tables
• Various tables used to specify the representation
  of data in the object models.

28
Entities as attributes

• Low cohesion
• Performance?

Object Class Structure Table FactoryF1(P)
has
Attribute Table TypeStart TimeStop Time
Enumerated Datatype Table PartEntityType
is enumerated
29
ER as an attribute

• High cohesion
• Performance?

Object Class Structure Table FactoryF1(P)
Fixed Record Datatype Table PartEntity TypeStart
TimeStop Time
has a
is of type
Attribute Table Part
is enumerated
Enumerated Datatype Table PartEntityType
30
ER as an interaction

• High cohesion
• Performance?

Interaction Class Structure Table ProducePart(P)
Fixed Record Datatype Table PartEntity TypeStart
TimeStop Time
has a
is of type
Parameter Table Part
is enumerated
Enumerated Datatype Table PartEntityType