SEIT: Modeling Radio Network Connectivity in a RealTime Distributed Simulation

1
SEIT Modeling Radio Network Connectivity in a
Real-Time Distributed Simulation

• By
• Peter Valentine, EPG
• Ari Nguyen, EPG

2
Introduction

• Focus How a degree of Communications Reality
  can be injected into a heterogeneous distributed
  simulation.
• Approach Attach simulated radios to virtual
  entities generated by the combat models
  (ONESAF/JANUS) and simulate the connectivity
  between members of radio networks.
• Distribute Connectivity tables to the other
  players in the simulation via custom DIS PDUs
• Modeling of Radio Network Connectivity was
  performed by EPGs custom simulation software
  package known as the Orion Electro-Magnetic
  Engineering Workbench (EMEW)

3
Network Architecture
Entity State
HLA
HLA
DREN
Entity State
Entity State
MAK HLA/DIS
RFNet Conn
HLA
Gateway
Entity State
RFNet Conn
Orion/EMEW
HLA RTI
DIS PDUs
4
Orion Software Architecture

• MissionSpace contains Virtual Entities of
  VEntities
• DIS Gateway imports selected DIS Entities as
  Transceiver VEntities into the MissionSpace
• RFNet VEntities are collections of Transceiver
  Members
• Member status is calculated automatically,
  orphaned members are identified.

5
DIS Gateway

• Imports Exports Entities via DIS PDUs
• Supports custom plug-in Mediators (SEIT
  Mediator)
• Reads jointly defined XML file to bind Radios
  Nets to URNs in DIS Marking fields
• Automatically sends a RFNet Connectivity Data PDU
  every 10 sec
• Draws radio characteristics from standard Orion
  Palettes

Sample screen of DIS EntityState PDUs driving
entities in the MissionSpace
6
WebMonitor Gateway

• A plug-in gateway that allows users to use a
  standard web browser to access any Orion/EMEW
  instance
• Turns Orion/EMEW into a WebServer
• Allowed remote participants to monitor Orion/EMEW
  from any convenient computer (platform
  independent)
• User could view the map, browse all active
  queries and inspect the properties of all
  VEntities

7
Orion/EMEW RF Modeling

• Full 3-D Environment
• Uses Terrain Integrated Rough Earth Model (TIREM)
• General Purpose Models of
• Transceivers
• 3D Antenna Patterns
• Communications Links
• Transmitter Footprint (RF Spider)
• Combined Area Coverage and optimal
  repeater/jammer locations for multiple
  Transceivers
• Dynamic environment, models automatically
  recalculate when either the user or other
  simulations change the properties or position of
  VEntities
• Dynamic displays results in 2D Map Views, 3D
  Views, and Tabular Grid Views

8
Modeling RF Networks

• Needed to model two basic kinds of Radio
  Networks
• Direct Connected Radios must be able to connect
  directly to the Network Controller, example
  Repeater Networks
• Peer-Routed Radios can forward packets through
  peers in the network until they reach the Network
  Controller, example EPLRS and other packet
  forwarding radio systems.
• Networks must be evaluated in real-time,
  specifically
• A number of networks needed to be evaluated
  throughout exercise/simulation (in IOC3 we
  modeled 17 networks)
• Networks can include a large number of members
  (in IOC3, several networks included 50 members)
• In large networks, the network topology can be
  changing at up to 10 times per second or more (50
  members sending position updates every 5 sec)
• We needed to fully evaluate all networks and
  provide a connectivity matrix for each network
  every 10 seconds

9
Direct Connected

• Direct Connected members only need to talk to
  the Net Controller in order to be connected.
• Failure to directly reach the Net Controller
  means that the member is orphaned from the
  network and cannot communicate with the other
  members
• Direct Connected RFNets are computationally
  efficient, each member need only evaluate a
  single link

Mbr 1
Mbr 2
Net Ctrl
Mbr 3
Mbr 5
10
Peer-Routed

• Messages can be routed across any number of
  connections to reach the Net Controller if the
  Net Controller cannot be reached directly due to
  terrain interference.
• The number of possible paths that must be
  evaluated in order to determine connectivity
  increases geometrically with the number of
  members in the network.

Net Ctrl
Mbr 1
Mbr 2
Mbr 3
Mbr 5
11
Impact of Network Architecture

• Network Traffic was very bursty due to
  interactions between the WAN and DIS/HLA
  Gateways.
• Increased packet loss required longer Entity
  Time-Outs
• Bursts were peaking at 800-1000 entity-state
  updates per second and overloading the
  Simulations ability to recalculate the RFNets

DIS Monitor built to monitor Remote Traffic

• By evaluating the bursty nature of the DIS
  Traffic at the remote sites, we were able to
  leverage the nature of the Traffic to improve
  model performance.
• Orion was designed to immediately recalculate the
  RFNet whenever any of its members changed its
  topology. By taking advantage of the bursts, we
  paused calculations during the burst, and only
  recalculated the network in time to broadcast our
  connectivity tables thus significantly reducing
  processor load.

12
Summary

• Modeling RFNets in real-time over a distributed
  network is an exercise in balance
• Must trade off accuracy vs performance
• Network performance has a critical impact on
  overall system performance
• It is possible to take detrimental network
  characteristics and turn them to your advantage
  by tweaking your models to take advantage of
  the nature of the traffic.