SBIR-STTR Award

High-Fidelity Network Simulation Using a Flexible Agent-Based Parallel Discrete Event Simulation Framework
Award last edited on: 2/1/2013

Sponsored Program
SBIR
Awarding Agency
DOD : Army
Total Award Amount
$799,944
Award Phase
2
Solicitation Topic Code
A09-171
Principal Investigator
Wilbur Peng

Company Information

Intelligent Automation Inc (AKA: IAI)

15400 Calhoun Drive Suite 190
Rockville, MD 20855
   (301) 294-5200
   contact@i-a-i.com
   www.i-a-i.com
Location: Single
Congr. District: 06
County: Montgomery

Phase I

Contract Number: ----------
Start Date: ----    Completed: ----
Phase I year
2010
Phase I Amount
$70,000
Highly parallel computing clusters hold out significant promise for improving fidelity of mobile ad-hoc network (MANET) simulations as well as increases in the scale of network simulations to hundreds or thousands of network nodes. However, harnessing such systems for parallel discrete event simulation (PDES) of MANETs presents significant challenges in model programming and implementation, achieving efficient processor utilization, and physical layer modeling. In this effort, we propose an agent-based simulation framework that incorporates the following aspects: 1) Flexible time management. We hypothesize that selective relaxation of some certain time management constraints can lead to significant performance increases while remaining within bounds of acceptable fidelity, 2) Reuse and integration of existing and new network models. An agent based approach will provide effective ways to integrate existing and new models within a PDES system, and 3) High fidelity physical layer simulation of carrier recovery, synchronization, demodulation, and decoding, and signal propagation. The result of the overall effort will provide a new high-fidelity scalable MANET simulation capability infeasible on current hardware.

Keywords:
Simulation, Networks, Discrete Event Simulation, Agents, Manet

Phase II

Contract Number: ----------
Start Date: ----    Completed: ----
Phase II year
2012
Phase II Amount
$729,944
High-fidelity simulation of large mobile ad-hoc networks consisting of hundreds to thousands of nodes can substantially facilitate development and deployment of network centric systems. However, current parallel discrete event simulation face scalability limitations and cannot effective utilize large numbers of processors for certain classes of simulation workloads. In this Phase II effort, an agent-based parallel discrete event simulation and modeling framework called Cynx will be developed that has the following key innovations: 1. An application programming interfaces that enables models to help the simulation framework to increase parallelism through improved estimation of lower bound on timestamp (LBTS) as well as topology-aware computation of agent lookahead values, 2. A simulation framework optimized for both shared memory multiprocessing that reduce communications and synchronization overhead between simulation agents, 3) Interfaces to high-fidelity MAC and physical layer simulation/emulation modules that enable integration of PDES systems with specialized computation capabilities and 4) A library of reusable components optimized for the Cynx framework demonstrating how models can be written to improve overall simulation scalability. The result of the proposed effort will be an open architecture and infrastructure for developing and executing large-scale MANET and networking simulations that will harness multicore computation clusters.

Keywords:
Parallel Discrete Event Simulation, Agent Programming, Multicore, Distributed Systems, Network Simulation