Dynamic Terrain: From RunTime Modifications to PreExercise Tailoring

1
Dynamic Terrain From Run-Time Modifications to
Pre-Exercise Tailoring

• David Bakeman
• Dr. Dale D. Miller
• Dr. Steve Adelson
• Kent Cauble

Lockheed Martin Information Systems Advanced
Simulation Center 3605 132nd Avenue
Southeast Bellevue, WA 98006
2
J9901 Overview

• Simulation experiment dealing with attack
  operations against critical mobile targets
• discovery, not demonstration
• theater ballistic missiles (TBM), transporter
  erector launchers (TEL) and their infrastructure
• Integration of STOW with Simulation of the
  Location and Attack of Mobile Enemy Missiles
  (SLAMEM) model (Toyon Research Corp.)
• Weapons systems circa 2015
• See Ceranowicz et al., J9901 Federation
  Development for Joint Experimentation, 99F-SIW-120

3
Requirements for Massive Feature Additions

• STOW SW Asia TDB selected as venue for experiment
• 5 X 7 degrees
• Feature data derived from ITD, PITD, WVS, DCW, et
  al.
• Strangely enough, there were
• insufficient forested and urban areas for hiding
  TELs
• insufficiently dense road networks to hide TELs
  and support vehicles among civilian traffic
• Requirements
• over 2,000 sq. km. of urban areas providing three
  levels of concealment
• nearly 50,000 sq. km. of forested areas
• over 5,000 km. of new roads.

4
Intensification Options

• Traditional approach
• back to the Data Base Generation System (DBGS),
  make edits, compile (over 30 hours), test,
  iterate
• Pre-exercise tailoring using Dynamic Terrain and
  Objects (DTO) technology
• edits distributed over the network with changes
  made directly to the run-time data bases
  (JointSAF CTDB, OpenScene ModStealth, DTO data
  bases)

5
DTO Architecture Overview
6
ECN Fragmentation
7
Technology Advancements for DTO

• New paradigm in terrain data base editing
• particularly useful when edits do not represent
  ground truth
• GIS produced ShapeFiles imported by DTSim
• Run-time canopies
• Urban area abstracts
• Run-time cut and filled roads
• including calculation of new road topology
• Saving of run-time data bases

8
Use of GIS for Definition of New Features

• For STOW ACTD
• DTO changes were made via
• JointSAF Obstacle Editor
• DTSim GUI
• developed as diagnostic tool only
• For ExInit purposes, ECNs logged to disk during
  offline SNE tailoring, distributed during
  exercise startup
• J9901 advancement
• Use ESRIs ArcView for creating and editing
  features
• Output to ShapeFiles
• Extended DTSim to import ShapeFiles

9
ArcView Images of SWA Data Base Before and After
the Creation of the New Roads, Canopies, and
Urban Areas.
10
JointSAF Plan View Display Before and After the
DTO Changes.
11
Import of GIS Data by DTSim

• ShapeFiles store geometry and attribution in a
  simple binary format
• get-next style API developed and distributed
  with S1000 API
• geodetic coordinates
• no limitations in spatial extents
• Four types implemented and imported by DTSim
• road linear geometry
• canopy areal geometry
• urban areal geometry
• multi-state object point geometry

12
Forest Canopy Representation

• JointSAF
• Previously, canopies represented in JSAF via
  Abstract and Physical Features
• Abstract Vector outline with attributes
• Physical canopy top polygons to support
  intervisibility from above
• Large area of canopies required would
  significantly increase CTDB size if polygonal
  representation were retained
• Represented as abstract feature only, with
  attribution driving detection of targets by
  overhead sensors
• OpenScene ModStealth
• Represented polygonally with canopy top and
  treeline side polygons
• Terrain polygons intersecting areal feature are
  duplicated, clipped, raised to canopy top, and
  retextured
• Treeline side polygons generated and textured

13
Urban Area Representation

• Existing data base had buildings and other 3-D
  models in major urban areas
• but not of real world density
• JSAF required areal abstraction similar to canopy
  abstract features
• urban center impenetrable to all sensors
• intermediate urban center degrading sensor
  detection by 70
• urban outskirt degrading sensor detection by 30
• Visual representation similar to canopies using
  textures for appropriate visual cues

14
Urban Area Representation
15
Attributes

• Roads

• Urban Areas

• Forest Canopies

16
Cut and Filled Roads

• Specification
• centerline geometry, width, texture, STGS,
  transition zone width
• Intersection with existing terrain
• generates 3-D profile along the roadbed
• Construction
• expand 3-D centerline to create zero-roll road
  polygons
• Integration
• generate transition zone polygons to sew the
  roads into the existing terrain
• Emplacement
• clip out old terrain polygons with polygon
  delete and add-child ECNs
• add new road polygons with polygon add ECN

17
Construction of Road Polygons from the Centerline
18
Limitations

• Intelligent road placement is solely at the
  discretion of the user
• Curves greater than 90 degrees are illegal
• Roads are flat relative to the underlying terrain
• no raised roadbeds
• problematicat intersections
• no banking
• Integration zone is small, and cannot be used as
  a shoulder

19
OpenScene ModStealth View of a Section of Newly
Created Road
20
Saving to Disk

• For the STOW ACTD, SNE tailoring was performed by
  having the DTScribe play back a log of ECNs from
  disk
• Several hundred changes (survivability positions,
  tank ditches, buildings) were performed in about
  a minute
• Massive changes required for J9901 would have
  required hours
• Three applications extended to save their
  run-time data bases to disk
• JointSAF CTDB
• OpenScene ModStealth GDE
• DTSim S1000
• These modified data bases were cut to CD and
  predistributed as usual

21
Timing

• Total run time for DTSim to create all Urban
  Areas, Canopies and Roads 7.5 hours
• CTDB save to disk (including recalculation of all
  road topology) 1.5 hours
• conservative estimate of road topology
  calculation of 400 m/sec.
• S1000 save to disk 2 hours
• Visual GDE save to disk
• 3 hours save
• 3.5 hours to pass the data through the stealth to
  IG TCP/IP interface
• TOTAL 17 hours
• or about half the time of simply compiling the
  CTDB from S1000

22
Robustification of Software

• Implementation of Disk Based Memory Caches
• Previous DTO exercises totaled 3 MB ECN data
• J9901 over 330 MB ECN data
• Improvement to the DTSim to Scribe ECN Request
  Protocol
• Single ECNs as large as 30 MB, and original
  fragmentation scheme failed for ECN requests
• Creation of Synchronization between SIM and
  SCRIBE for ECN Requests
• Increase Data Transmission Sizes for Stealth to
  IG Communication
• Implementation of Disk Based Memory Cache for IG
• Network Difficulties
• RTI-s leaking 100 - 200 MB for each ECN
• Not fixed DIS mode used instead