Parallel Simulation

1
Parallel Simulation

• By Sameer Hussain, 1727466
• Benaceur Assif, 1871157
• Date Tuesday, March 12, 2002
• This report was prepared for Professor L.
  Orozco-Barbosa in partial fulfillment of the
  requirements for the course ELG/CEG 4183.

2
Topics

• Simulation
• Types of Simulation
• Parallel Simulation
• PARSEC, PAVE, Compose.
• Future of Parallel Simulation
• Questions
• Bibliography

3
What is simulation?

• Definition 1 The use of mathematical/logical
  model as a way of answering question about a
  particular system.
• Definition 2 Computer simulation is the
  discipline of designing a model of an actual or
  theoretical physical system, executing the model
  on a digital computer, and analyzing the
  execution output.

4
Why is simulation important?

• Simulation is often essential when
• The model is very complex with many variables and
  interacting components.
• the underlying variables relationships are
  nonlinear
• the model contains random variables
• the model output is to be visual as in a 3D
  computer animation.

5
Use of Simulation

• Typically simulation is used for planning and
  design.
• Telecommunication networks
• Transportation systems
• Logistics
• Electronic systems (for microelectronics,
  computer systems)
• Battlefield simulation (red army vs. blue army)
• Manufacturing systems
• Ecological systems.

6
System Dynamics vs. Discrete Event

• Simulations can be event based or time based.
• System Dynamic
• Analysis in a continuous time period
• Looks at systems at a very high level. More
  suited for decision making
• Effect shipping delay time has on production,
  inventory, sales
• Discrete Event
• Analysis in a specific time horizon
• Looks at system in a predictive manner
• How many resources do we need to achieve a
  certain throughput?

7
Types of Simulation

• There are three types of computers simulations
• Sequential Events are schedules one after
  another on a single processor.
• Parallel multiprocessor or cluster of
  workstations on a LAN
• Distributed Workstation and virtual simulators
  on a WAN

8
Comparison between simulations

• Sequential vs. Parallel Simulation
• Some systems are too complex for sequential
  simulation
• Parallel vs. Distributed Simulation
• Distributed Simulations are too complex to design
  and control.
• A lack of good existing tools.
• Parallel simulation can be used to model complex
  systems while still being relatively easy to
  design. This is why it is very popular with
  researchers.

9
Parallel Simulation

• Parallel simulation always deals with discrete
  event systems.
• Advantages of parallel simulation
• Reduced model execution time
• Scalable performance
• Geographically distributed users and/or resources
  
• Integrate simulations running on different
  platforms
• Fault tolerance

10
PARSEC

• PARSEC stands for PARallel Simulation Environment
  for Complex System
• It is a C-based simulation language derived from
  Maisie. Developed by the Parallel Computing
  Laboratory at UCLA
• It can be used for sequential and parallel
  execution of discrete-event simulation models.
• PARSEC programs consist of entities, which
  exchange messages.

11
PARSEC Layer
12
Use of PARSEC

• In Wireless
• TCP Performance for Multi-hop Networks
• AMRIS A Multicast Protocol for Ad hoc
• Providing Multiple Service Classes for Bursty
  Data Traffic in Cellular Networks
• Multi-hop Protocols
• Simulation and Analysis of the Cache Group
  Management Protocol
• Architectural Models for Avionics Communications
  Systems

13
PAVE and Compose

• PAVE
• PAVE stands for Parsec Visual Environment
• It is used to support visual hierarchical
  design of simulation models.
• Compose
• C library that can be interfaced with native
  C to execute parallel simulations.

14
UCLA Simulation Environment
15
GloMoSim Architecture

• It uses the parallel discrete-event simulation
  capacity provided by Parsec. It currently
  supports protocols for a purely wireless network.

16
Future

• Short Term
• GloMoSim will have functions for supporting wired
  networks and hybrid network (with both wired and
  wireless capabilities)
• Long Term
• Distributed Simulation will replace parallel
  simulation once the proper tools are developed.
• More System Dynamic Simulation instead of
  Discrete Event simulation.

17
Questions

• What is simulation and why must we use it?
• Simulation is using logical/mathematical models
  to study and solve physical systems.
• The model is very complex with many variables and
  interacting components.. The variables are
  nonlinear. The model may contain random variables
  and the output has to be visual.
• What is the difference between discrete event
  simulation and system dynamic?
• Discrete event look at the system in a predictive
  manner and and analysis is for a specific time
  period
• System dynamic is more suited for decision making
  and is performed for continuous time.

18
Questions

• What are the types of computer simulation and
  which one is currently popular?
• Sequential simulation, parallel simulation,
  distributed simulation.
• Parallel simulation is popular since it can model
  more complex systems than sequential simulation
  while not being as complex to manage and program
  as distributed systems.
• Why is Parsec so important? What tools can be
  used with it?
• It allows parallel simulation.
• PAVE and Compose can be used with PARSEC. In
  addition, GloMoSim uses PARSEC calls to simulate
  wireless networks

19
Questions

• What can be modeled using parallel simulation?
• Various discrete event systems such as
• ATM
• Wireless Networks
• Avanoics Communication systems
• Mobile IP

20
Bibliography

• Arsham H., The Use of Simulation in Discrete
  Event Dynamic Systems Design, Journal of System
  Science, 31(5), 563-573, 2000.
• R. Bagrodia, R. Meyer, M. Takai, Y. Chen, X.
  Zeng, J. Martin, and H.Y. Song, "PARSEC A
  Parallel Simulation Environment for Complex
  Systems, " IEEE Computer, vol. 31, no. 10,
  October 1998, pp.77-85.
• R. Bagrodia and M. Gerla, "A Modular and Scalable
  Simulation Tool for Large Wireless Networks",
  International Conference on Modeling Techniques
  and Tools for Computer Performance Evaluation,
  1998.
• Erik Dirkx, Parallel Computer Networks
  Simulations An Evaluation, Proceedings of
  European Simulation MultiConference, pp. 713-716,
  Nuerenberg, June 1990.
• Xiang Zeng, GloMoSim a Library for Parallel
  Simulation of Large-scale Wireless Networks,
  Proceedings of the 12th Workshop on Parallel and
  Distributed Simulations -- PADS '98, May 26-29,
  1998 in Banff, Alberta, Canada.