Simulation Interoperability Toolkit

1
Simulation Interoperability Toolkit

• James Nutaro and Mike Kamrowski, Raytheon
• Bernard Zeigler and Hessam Sarjoughian, ACIMS
• Aug. 2001

2
SIT

• SIT Simulation Interoperability Toolkit
• SIT is a set of tools that supports the
  construction of distributed simulations
• Simulations can be live or constructive
  exercises
• Human in the loop simulations, hardware in the
  loop/computer in the loop simulations, live data
  sources
• Engineering trade studies, CONOPS, high fidelity
  engineering simulations

3
SIT

• Consists of four tools
• Tapir - A portable API specification for
  distributed simulation middleware
• Federation Bridge - A middleware bridging tool
• Embedded RTI - High performance, small footprint
  message passing system supporting the Tapir
  interface
• Fit - The FOM interface tool

4
Tapir

• Provides transparent access to common simulation
  infrastructures
• HLA, TENA
• API specification is minimal and complete
• Readily implemented with low level toolkits
  such as MPI and CORBA
• Synchronization protocols for constructive and
  live simulation of DEVS models
• Common distributed programming tools
• Barrier synchronization
• Push and pull communication models

5
Federation bridge

• For live exercises, supports infrastructure
  bridge
• HLA/HLA
• HLA/TENA
• Bridge wire standard for injecting and/or
  extracting data from a running live exercise

6
Embedded RTI

• High performance, portable, small footprint
  message passing system
• Supports the Tapir API
• For live exercises, can be interfaced with
  other middleware systems via the Federation bridge

7
Three tiered view of distributed simulation
SIT

• SIT bridges the gap between local simulations and
  the distributed simulation system that connects
  them
• Tapir supports a one middleware model for
  distributed simulation exercises
• Bridge supports a many middleware model for
  distributed simulation

8
Many middleware model
Embedded RTI
DMSO RTI
Federation bridge
Many, disparate middlewares connected by a bridge
TENA
9
Many middleware model
Embedded RTI
DMSO RTI
Federation bridge
Appears to federates as...
TENA
10
Many middleware model
A single middleware system!
11
Enabling technology

• The federation bridge is the enabling technology
  for the many middleware model
• Advantages include
• Cheap and easy interoperability since individuals
  federates do not need to change
• Federation developers can select middleware to
  suit local needs, within limits!
• Allows a third party to treat federates and whole
  federations as black boxes to be hooked together
• Limitations
• Limited to message passing between federates
• Advanced middleware features such as logical
  clocks and QoS guarantees can not be transferred
  across a bridge

12
one middleware model
native API
native API
Standard API
Standard API
Standard API
13
Enabling technology

• The Tapir API specification
• Some advantages of the one middleware model
• Access to native middleware services via the
  standard API
• for example barrier synchronization, logical
  clocks
• Simpler distributed applications (less software,
  fewer things to go wrong)
• Some disadvantages
• Interoperability relies on middleware
  independence, federates must be constructed with
  middleware independence in mind
• Likely to require source code changes for legacy
  software (for example, HLA only or DIS only
  simulations)

14
Fit FOM Interface Tool

• Generates C wrapper classes around Tapirs
  typed, parameterized messages
• Wrapper code is automatically generated from a
  DIF formatted OMT file

15
Current Status

• A working example of SIT has been constructed
  using the HLA
• To integrate a federate, the simulation developer
  builds an interface map that translates internal
  simulation data to/from the middlewares typed,
  parameterized messages.
• The developer includes a handful of API
  middleware calls in the simulation code
• Potential port to TENA, a range interoperability
  toolkit for constructing virtual proving grounds.
  

16
Future work

• Development of a high-performance Tapir
  implementation for small scale distributed and
  parallel simulations
• Continuing evolution of the API specification to
  ensure portability of concepts