High Performance Computing in support of Joint Urban Operations

1
High Performance Computingin support ofJoint
Urban Operations

• Bob Lucas and Dan Davis
• USC/ISI

2
OUTLINE

• Urban Resolve Overview
• ISI Contribution

3
Title Page, CPLAN 01 Overview, 11 Sep 00
Modeling and Simulation Support To Joint Urban
Operations Experiment URBAN RESOLVE
4
Urban Resolve Experiment Concept

• Urban Resolve is a three - phased experiment with
  each phase consisting of the following elements
• Wargames and Workshops
• Constructive Simulations
• Human-In-The-Loop (HITL) Simulations
• Focus is on exploring three fundamental
  capabilities which are key to the success of
  future JUO concepts
• 1. The capability to achieve a high-level of
  situational understanding
• 2. The capability to shape the urban battlespace
  through precision strikes against targets in
  urban areas
• 3. The capability to engage by maneuvering joint
  forces in the urban environment

5
Multi-Resolution Synthetic Environment
6
(No Transcript)
7
STEALTH View-Urban Area
8
Urban Lots of People

• Realistic Urban clutter Civilian vehicles,
  people,
• Demonstrators, protestors, men, women, children,
  
• Life-like human characters respond to real-time
  simulated scenarios in high-resolution
  environment
• Move realistically, respond to simple commands,
• Demographically correct, move in environment as
  directed
• Respond to real-time commands to change activity

9
Stealth View of Clutter Crowd
10
Sensor/Platform Architecture
20000m
Hyperspectral EO/MWIR
5000m
Ku-Band SAR/GMTI 3D Laser Profiler
HS, EO/MWIR, 3D Ladar
Red Spike/Blue Spike EO/MWIR
FOPEN SAR Elint
EO/LWIR, RF/DF, UGS Emplacement
EO/LWIR 3D Ladar Magnetometer
FOPEN SAR, Elint EO/LWIR
X-Ray
0 1000m
UGS
Humint w. AGS
On the Ground
TAGS
11
SLAMEM Sensor Federate
Provides Platforms Sensors for HITL and
Constructive Trials
JSAF PVD
JSTARS
12
The Scalability Challenge ()
2,000,000
Future experiments require orders of magnitude
larger more complex battlespaces
Got to do SCALE and FIDELITY
Number and Complexity of JSAF Entities
1,000,000
SPP Proof of Principle DARPA / CalTech
120,000
107,000
50,000
12,000
3,600
UE 98-1 (1997)
J9901 (1999)
SAF Express (1997)
JSAF/SPP Goal (2006)
AO-00 (2000)
JSAF/SPP Urban Resolve (2004)
JSAF/SPP Tests (2004)
13
ISI Contributions

• HPCMP Distributed Center
• Scalable Communications
• Data Logging
• Data Management

14
Prior Art SFExpress
DARPA/SPAWAR Caltech/JPL
100,298 modSAF entities Used all HPCMP MSRCs
15
HPCMP Distributed Center

• ISI wrote JFCOM proposal
• Deployed, spring 04
• ISI Manages for JFCOM
• DREN Connectivity
• Example of real-time, interactive supercomputing
• Application tolerates network latency

16
Scalable Router Architecture
17
Router Performance Comparison
Warning Blunt Crayon Smoothing Of Raw Data
18
Experimental Data Collection

• Need to record every visible action taken by an
  entity
• Even if nobody else sees it
• Implemented new logger
• Key feature is interceptors
• Displaced HLA results
• Raw data stored locally
• Allows near real time queries

19
Data Management

• Analysts make relational queries
• After-action
• Brought SOA database to JFCOM
• Replaced MS Access (32-bit)
• Using SQL-Lite
• ISI has taken lead in managing all data archiving
  as well as analysis tools
• J9 experiments generate TBytes of data
• Three order-of-magnitude increase from 02
• Could easily increase another 100X

20
GPU Feasibility Experiment

• DARPA IPTO PCA project
• Characterize Line-Of-Sight bottleneck in Urban
  Resolve
• Can Graphics Processors alleviate bottleneck?
• Leverage UNC work with Armys oneSAF

21
Summary

• JFCOM J9s JUO is a huge success
• DODs most impressive synthetic environment
• JSAF could still use lots of enhancement
• Doesnt accurately model communications
• ISI Technology is critical
• Scalability
• Data Collection and Management