Communication Networks – Modeling, Simulation & · PDF fileCommunication Networks...
Transcript of Communication Networks – Modeling, Simulation & · PDF fileCommunication Networks...
Communication Networks –Modeling, Simulation & Emulation
Jim Nutaro
Computational Sciences and Engineering Division
Goal - Modeling of Large Scale, Complex System ofSystems that are Dependent on Information Flow
• Future Combat System
– ORNL part of FIST team.
• Real-time Cyber SecuritySituational Awareness
– Joint USSTRATCOM-DISAEffort for DOD.
– ORNL lead player forUSSTRATCOM.
• Critical InfrastructureNetwork
– Example: The Electric PowerGrid
– ORNL DHS National Lab incharge of TVTA for cyber.
• Sensor Networks
• Information Operations
Metropolitan Area
Network (MAN)Integrated Space and Terrestrial Wide Area Network
MAN
Assured
Net Cached Data Regional and Globally
Distributed Processing
Collaboration
Services
Secure IP Network
Convergence Layer
Battlespace Reach back
and Smart Pull61 el. T
heater
8GHz MBR
169 el. 7GHz M
BT44
/
20
256 element
TPA484 element
RPA
Data
Data
Data
Data
MANT-SAT
T-SAT
T-SAT
T-SAT
Communications M&S – Overview
Interoperability- Wired, wireless, hybrid simulators
- Simulator-independent models
Applications- Network security
- Testing and analysis
Modeling- Wireless for MOUT
- Modeling frameworks
Visualization- 104-106 nodes, links
- Animation & Sonification
Scalability- Size and speed
- Supercomputing
Interoperable Tools – Example Scenario Ship-to-ship, ship-to-shore, ship-to-stationnetwork
• 7 Ships
• 1 Ground station
• 1 Satellite - 220ms latency, 512kbpspeak shared bandwidth
• Packeteer traffic shaper on each ship
• Ethernet LAN on each ship
• COTS hardware/software on each ship
• Running Windows NT, Solaris
• Lotus Domino, Internet Explorer, …
On-shore &airborne mobilead-hoc networks
Given:
• Afloat Naval Exercise Data (JWID00) atReconfigurable Land-based Test Site (RLBTS)modeled in ns-2
•Ship-to-shore & Shore-to-ObjectiveManeuver (STOM) data modeled in GloMoSim
GloMoSimPDNS
Integrated Tool Configuration
satellite link Line-of-sight link
• Intra-ship LAN traffic - File transfer, web & email proxies, …
• Inter-ship WAN traffic - Position updates, calls for fire, … - Web, email, chat, …
• Cluster traffic - Images, orders, updates, …
• Remote traffic, … - Orders, video streams, …
Ship
Ship Ship
Ship
Mobilenodes
Integration Methodology & Framework
Runtime Infrastructure (RTI)
Integration Services
• Semantic Integration
– Protocol and item registration
– Consensus computation
– Message Export / Import services
• Parallel Execution
– Multicast-based Event distribution
– Synchronization (Time Management)
Parallel/ Distributed
Simulation Protocol
DEVS
Joint
Measure
A platform effectivenessmodeling toolCreated at LockheedMartin.
Models are specified usingthe DEVS formalism.
DEVS is implemented as anobject oriented simulationtoolkit.
Joint MEASURE wasimplemented with DEVS-C++ from the University ofArizona.
Example – DEVS & Support of Joint Measure
xxxCommand
ControlModel
xxxxLaserModel
x
x
x
SpaceBased
Discrimination
x
x
x
SpaceBasedLaser
x
x
x
x
x
x
x
x
x
MissileDefense
(Theater /National)
xxxxxxWaypoint& HeadingNav Model
xxxxOrbital
PropagateModel
x
x
x
Integrated System
Center
x
x
x
CommonAero
Vehicle
x
x
x
x
JointCompositeTrackingNetwork
xxBallistic
TrajectoryModel
xWeatherModel
xxxxEarth &TerrainModel
xxComm.
Model
xMissileModel
xxIR SensorModel
xxxxRadarModel
SpaceOperations
Vehicle
Coast GuardDeep Water
ArsenalShip
GlobalPositioningSystem III
CriticalMobileTarget
Project
Model
Component Model (Actual Reuse) Matrix
Parallel/Distributed Network Simulation
ns-2Sequential
Simulator
ns-2Sequential
Simulator
ns-2Sequential
Simulator
ns-2Sequential
Simulator
Runtime Infrastructure Other
Services
Virtual Time
Synchronization
Asynchronous
Distributed
Reductions
Heterogeneous Network
Communication
Runtime Interface
Application
(GTNetS, ns-2, …)
Popular sequential models (ns-2)parallelized by integrating multipleinstances of itself.
Supercomputer
Network Models
Large-scale Packet-level Simulation
0
20
40
60
80
100
120
140
0 256 512 768 1024 1280 1536
Processors
Millio
n P
kt h
op
s/sec
Ideal/Linear
Target
Achieved
106 million
pkt hops/sec, 4 million
simulated nodes
Some of our recent milestones in parallel/distributed network simulations:• Record: Among the largest & fastest packet-level network simulations to date
• Shown: No. of packet transmissions simulated per second with pdns on Lemieux
• Benchmark: A “standard” benchmark (Campus network) scenario
Simulation Platform: Lemieux
Pittsburgh Supercomputing Centerwww.psc.edu/machines/tcs/lemieux.html
Reported inMASCOTS’03
Testing Network Defenses
• Interface actual securityapplications with virtualnetworks
– Transparent, plug-n-playinto large virtual “networkbattlefield”
– Test against large-scalevirtual attack scenarios
– Requires TCP/IPemulation…
Monitoring/
Detection/Analysis
Real
Large-scale Network & Malware
Virtual
Example - Honeypot Emulation
• Attack actual honeypot with simulated worms
• Test honeypot operation and effectiveness
Honeypot
PDNS
TCP/IP Emulation
Visualization of Multiple Campus Networks
10 campus
networks
connected
together
Portions of “Military Enterprise” Networks
dartmouth3886 ornl9177
campus538
*Network snapshots from netanim visualizer
Run
Run
Run
Fast radio channel simulation
Radio wave propagation through a gap in a concrete wall.
t = 3.27E-9 t = 5.80E-9 t = 1.09E-8
A comparison of empirical and wave propagation models
Transmitter
92.770.7Log-distance path loss,n = 4
76.259.7Log-distance path loss,n = 3
112.4948.7Free space path losswith knife edgediffraction
59.748.7Free space path loss
78.358.3Wave equation
RX 2 pathloss
RX 1 pathloss
Model
Receivers
RX 2
RX 1
TX
RX 1TX
RX 2
In Conclusion
• ORNL has a substantial research program in large scalesystem modeling, with an emphasis on communicationsdependent systems
• Simulation integration– Component based simulation
– DoD High Level Architecture
• High performance computing for large scale systems– Parallel discrete event simulation
– High performance simulation with component based frameworks
• Communication Network Applications– M&S for network security
– Network emulation
– Wireless channel modeling
– Visualization and Sonification
• Modeling frameworks for complex systems