SBIR Phase II Adaptive Collaborative Command Briefing

1
SBIR Phase II Adaptive Collaborative Command
Briefing
C2insight A CPOF-Compatible Commanders Tool for
Enhanced Battlespace Understanding and Adaptation

DARPA Sponsor J. Christopher Ramming
Kathleen M. Carley Jeff Reminga, Il-Chul Moon,
Mike Schneider
Roger Chapman, Don Benton, Floyd Glenn, Andrew
Rosoff
Joint Experimentation Directorate (J9)Aug 31st
2006
2
Presentation Overview

• Brief who we are and what we do
• C2insight Project Overview
• Phase I Pattern Discovery in CPoF repositories
• Phase II Build C2insight application for C2
  users
• Relevance to research at J9 and potential to
  integrate our efforts
• Our initial CPoF work
• CPoF data extraction analysis
• Related survey data collection analysis
• Summary of results
• Potential to leverage in a near-term
  collaborative effort with J9
• Our C2insight Phase II work
• The development plan for C2insight
• Integration with Urban Resolve 2015 in further
  work with J9

3
CHI Systems, Inc.
Creating solutions that help people, processes,
and products reach their fullest potential
through the application of user-driven technology.

• CHI Systems is a research, development, and
  engineering company that
• Creates solutions that humanize advanced
  technologies
• Makes complex technology more usable, intuitive,
  and effective for individuals and organizations
• Works with clients to help them use and adapt
  technology to improve work performance, mission
  success, and cost effectiveness.
• Merged with OSI Geospatial, Inc. in December 2005
• Software systems for real-time geospatial
  situational awareness
• 160 employees, 23M 2005 revenues
• 3 divisions
• OSI (Vancouver)
• CHI Systems, Inc. (Philadelphia, San Diego,
  Orlando)
• Mapcon Mapping, Inc. (Salt Lake City, Vancouver)

4
CHI Systems Primary Application Areas

• Decision-Support and Work-support Systems
• Command and control operations
• Visualization, collaboration, COA, SA
• Homeland defense
• Terrorist activity detection, emergency response
• Medical informatics and health record
  mobilization
• Mobile interfaces and aids
• Interactive electronic technical and flight
  manuals
• Technical, operations, maintenance, safety
• Training Systems
• Training requirements
• Simulation-based and game-based trainers
• Synthetic intelligent players, teammates and
  instructors
• Intelligent Tutoring Systems

5
Center for Computational Analysis of Social
Organizational Systems, Carnegie Mellon

• Focus combining computer science and social
  science to solve real world problems
• Methodological Areas of Specialization
• Dynamic and social network analysis
• Multi-agent systems
• Statistical techniques for data text mining
• Application areas
• Counter-terrorism, counter-narcotics, information
  assurance, OOTW, MMV, information diffusion,
  epidemiology, open-source, team behavior,
  organizational design, turnover, leadership, data
  privacy
• University wide center housed in the School of
  Computer Science
• 15 staff
• 12 Ph.D. students
• 3 faculty
• Sponsors and Clients
• NSF, NIH, DTRA, Darpa, ONR, AFOSR, ARI, ARL, DOD,
  misc. agencies
• Director Kathleen M. Carley kathleen.carley_at_cs.
  cmu.edu
• www.casos.cs.cmu.edu

6
CMU Background
7
Presentation Overview

• Brief who we are and what we do
• C2insight Project Overview
• Phase I Pattern Discovery in CPoF repositories
• Phase II Build C2insight application for C2
  users
• Relevance to research at J9 and potential to
  integrate our efforts
• Our initial CPoF work
• CPoF data extraction analysis
• Related survey data collection analysis
• Summary of results
• Potential to leverage in a near-term
  collaborative effort with J9
• Our C2insight Phase II work
• The development plan for C2insight
• Integration with Urban Resolve 2015 in further
  work with J9

8
Collaboration in CPoF
(http//mayaviz.com/web/industries/military/indus
try_mil_darpa_cpof.mtml) (http//www.oculusinfo.co
m/portfolio3.html)
Objects created are pedigreed
Distributed CPoF users can collaborate by sharing
workspaces
9
Objective and Technical Approach
Objective

• Utilize repository and other C2 data sources to
  provide feedback in CPoF and future
  network-centric C2 systems that facilitates
  adaptive command collaboration
• Feedback individual and group blue force
  (including command staff), blue on red, and red
  force (to some degree) activities
• - in context and augmented with analytical and
  visualization capabilities
• Adaptation command staff operations and
  operations in the battlespace

Use cases

• Briefings
• Pre-BUB and BUB automated reports for CG and
  others
• Presentations and question answering during
  briefings
• e.g. what are the red force patterns in this
  area?
• what actions appear to reduce attacks on
  us?
• Intelligence Analysis
• Is there evidence to support a tip that a police
  chief is corrupt?
• Planning
• Who should plan this operation in this area?
• Execution
• Have my brigade battle captains been updating
  their workspaces?

Approach

• Multidisciplinary
• Contextualized Social Network Theory for data
  analysis
• Human Factors for cognitive task analysis
  interface design and evaluation
• Software Engineering for data transformation
  management
• Utilization of exiting tools e.g. ORA, GIS
  software
• SME involvement in the engineering lifecycle
• Utilization enhancement of CPoF users work
  patterns

Broad spectrum SA enhancement for CPoF users from
live combat data in context
10
Innovation Value
Value
Innovation

• A reduction in the time taken to answer CG BUB
  CUB questions that require aggregating data
  available in the C2insight database
• An increase in the evidence used to support
  assessments presented to the CG, using data
  available in C2insight, and an increase in the
  confidence of the presenter
• An increase in the accuracy of the commanders and
  staff assessments of who the leading key actors
  are for important roles in this context, based on
  direct manipulation with C2insight and from
  customized and automatically generated reports
• CPoF users using C2insight indicate they believe
  it improves their efficiency and effectiveness

• Exploits a very large database of live combat
  data (most of which is already being recorded but
  not fully utilized).
• Models and offers this information to the user in
  a form which allows proven analysis methods and
  metrics to be applied, while also intrinsically
  supporting visual pattern exploration.
• Integration of geospatial and network information

E.g. Extracted CPoF data modeled as a network
workstation
event icon
abstract views coupled with geographic maps
provide insight
Mockup snapshot of a user exploring C2insight
interactive legend
geosticky icon
temporal context
proximity relationship
Network data In context
created by relationship
geographic context
filters
familiar frames of reference with direct
manipulation
The network data must be made meaningful and
manipulated naturally
11
Creators/Modifiers of items are clearly
identified along with their CPoF icons
Timeline of activity is displayed so that
specific periods can be investigated
Provides overviews
Items with a location are displayed both on the
map and along with their other relationships
Periods of interest are easily specified on the
timelines
Filters to focus attention
C2insight will have replay capability
C2insight will have replay capability
Items can be selected through direct manipulation
Early C2insight conceptual interface
12
Vision for work with J9

• Help evaluate impact of different organizational
  structures and get feedback on post operations
  value of C2insight
• Phase I Experiment at J9. Automatic and human
  data collection followed by visualizations and
  performance assessment of teams and individuals
• Phase II Participation in Urban Resolve 2015.
  Data collection and post operations analysis
• Evaluate real-time impact of C2insight
• Phase III Participation in Urban Resolve 2015.
  Data collection and real-time interaction with
  participants, followed by analysis

13
Presentation Overview

• Brief who we are and what we do
• C2insight Project Overview
• Phase I Pattern Discovery in CPoF repositories
• Phase II Build C2insight application for C2
  users
• Relevance to research at J9 and potential to
  integrate our efforts
• Our initial CPoF work
• CPoF data extraction analysis
• Related survey data collection analysis
• Summary of results
• Potential to leverage in a near-term
  collaborative effort with J9
• Our C2insight Phase II work
• The development plan for C2insight
• Integration with Urban Resolve 2015 in further
  work with J9

14
Dynamic Network Modeling Analysis of
Organizations
Social Network Analysis
15
CPoF Data Representation
CPoF Repository
User and battlespace activity is recorded by CPoF
automatically, in real-time
16
CPoF Data Representation
CPoF Repository

• Pedigreed information
• Events
• Tasks
• Geostickies

A lot of CPoF captured data is pedigreed,
revealing who has created/contributed to it and
when
17
Data Transformation Process
CPoF Repository
SME feedback
SME feedback
Filter and aggregate
Direct load into DB
DynamicNetwork
Translate, reschema
SME feedback
Software modules were developed to extract
specific data of interest from CPoF
18
Network Representationof CPoF Activity
This information was then modeled as a set of
networks of entities and relations
19
Network Representationof CPoF Activity
time
and repeated for different time periods
20
Summaries of Activity over Time
SMEs looking at structural aspects of these
networks, as well as their change over
time,found several types of patterns
interesting, but always wanted as much context as
possible
21
Patterns Explored

• Red Force Blue ForceInteraction Patterns
• Intra-organizational responsiveness to
  battlespace activityIn regard to
• novel team formation
• reactive planning

22
Patterns Explored

• Red Force Blue ForceInteraction Patterns
• Intra-organizational responsiveness to
  battlespace activityIn regard to
• novel team formation
• reactive planning
• Temporal Patterns
• Compared intensity and pattern of collaboration
  between, e.g.
• day shifts vs. night shifts
• Self-organizing andadaptation trends

Blue force and the interaction between blue and
red force patterns were explored
23
Patterns Explored

• Red Force Blue ForceInteraction Patterns
• Intra-organizational responsiveness to
  battlespace activityIn regard to
• novel team formation
• reactive planning
• Temporal Patterns
• Compared intensity and pattern of collaboration
  between, e.g.
• day shifts vs. night shifts
• Self-organizing andadaptation trends
• Infrastructure Relationships
• event geostickie proximity

Red force infrastructure patterns were also
explored
24
Ft Leavenworth Experiment

• Build 2 Objective to examine roles,
  responsibilities, tasks, products, structure and
  information requirements required of the UEx and
  Support Brigade staffs to conduct effective
  battle command.
• Build 2 Issues
• How does the UEx conduct operational level
  planning (mission orders, sequel planning)?
• How does the UEx synchronize internal C2
  operations?
• How does the UEx C2 Subordinate Brigade
  Operations?
• How do the Subordinate Brigades conduct dynamic
  deliberate branch planning?
• How do the Subordinate Brigades execute
  Supported/Supporting mission roles?

KUURDAMIR
ZARDOB
SABIRABAD
OBJ MIDDLE
IMISLI
PUSHKINO
OBJ BRAVO
OBJ BRAVO
This gave us access to unclassified CPoF data and
experience participating in C2 experiments
25
Organizational Tasks
Ft Leavenworth Experiment

• UEy
• Support execution of battle plan
• Coordination of Joint and SOF resources
• Support response to counter-attack as force
  provider
• UEx
• Execute battle plan
• UEx inter/intra-staff interactions (TACs, Main,
  MCG and UEy / Support Brigades)
• Execute parallel planning
• Integrated Operations (Stability Operations)
• Urban Operations
• Respond to counter-attack
• Reestablish crossing sites and LOCs
• Mobile strike / shaping fight
• Rear area threat
• Joint fires and effects coordination
• A2C2 coordination
• All Subordinate Brigades
• Support execution of battle plan
• Support response to counter-attack

• Sustainment Brigade
• Casualty Evacuation (BlueFor and Civilian)
• Mass Casualty Evacuation
• Movement control issues
• Rear area security issues
• Fires Brigade
• Support conduct Mobile strike.
• Interactions between former FSE staffs
• Replacement for the DOCC
• TCF commitment
• Rear area security issues
• A2C2 coordination
• Aviation Brigade
• Plan and conduct Mobile strike.
• Air assault (NGO MSEL) AVN BDE primary
  planners maneuver unit is force provider
• Support TCF commitment (reallocation of forces)
• Emergency resupply
• A2C2 coordination
• ME Brigade

The organization was large and distributed, and
representative of real-world C2
26
CMU Approach Combine DNA and Battle Lab Data in
Near Real Time

• Battle Lab data
• Collected data at Ft. Leavenworth unit of action
  experiment - this included questionnaire and CPOF
  data
• Analyzed both sets of data using ORAs DNA
  toolkit
• Provided feedback to members of the UA each day
  using DNA metrics
• Supported UA commanders in assessing
  organizational health of the UA during battle lab
  experiments
• Contrasted results from non CPOF data with
  earlier data collected by Dr. Gonzalez at other
  experiments
• Contrasted CPOF data with data from CPOF in Iraq

• Dynamic Network Analysis
• Examines relations among actors, resources,
  knowledge, tasks, etc.
• Graph theoretic metrics derived from social,
  organizational, operations research and
  psychological theory
• Combines social network theory, link analysis and
  multi-agent modeling
• Examine the constraints and enablers afforded by
  network position
• Examine change in these networks over time
• Utilization of DNA integrated tool suite
  particularly ORA

27
Leavenworth Dataset
Analysis of Events

• Three Repositories
• 2-5 Gb each
• 3 days of Activity
• No UCRs
• 25 Workstations
• 120 Role-players

id6
id7
id6
id7
We demonstrated we could provide feedback on who
created what type of events
28
CMU Leavenworth Dataset
Analysis of Tasks

• Three Repositories
• 2-5 Gb each
• 3 days of Activity
• No UCRs
• 25 Workstations
• 120 Role-players

id6
id7
id6
id7
CMU demonstrated we could provide feedback on who
created what type of tasks
29
CMU Illustrative Results
Identification of Key Actors
Mental Model Congruence
SWOT Trend Analysis
Emergent Sub-Group Identification
30
CMU Demonstrated Ability to Assess Over Time
Group Breaks and Rebuilds
Network Communications Data
Normal Operations
Splitting Operations
Split Operations Shared Situation Awareness
Shared SA Measure
Reunifying Operations Shared Situation Awareness
31
CMU Assessed Cliques at Leavenworth
Cliques
Note that the number of observed cliques
increases as the planning phase deadline
approaches (this in when order integration
occurs potential application point for AutoMap
32
CMU Assessed Emergent SubGroups

• Density, Knowledge, Resources, Tasks,
  Key-Leaders, Shared Situation Awareness,
  Boundary-Spanners, Cognitive Demand,
  Core-Periphery
• Example Shared Situation Awareness Combining
  questionnaire and CPOF data

33
Vision for Near-term work with J9

• CHI-CMU Provide
• Data Extraction Transformation
• Utilize existing code (may need some modification
  due to CPOF version differences)
• Expand implement a more comprehensive CPoF
  Metamatrix
• Develop support for additional data sources
• Visualizations and Analysis
• Imaging of behavior
• Over time trends
• Dynamic network analysis visualizations and
  metrics
• Geo-spatial-network information
• Blue force analysis (including ORA management
  report)

34
Presentation Overview

• Brief who we are and what we do
• C2insight Project Overview
• Phase I Pattern Discovery in CPoF repositories
• Phase II Build C2insight application for C2
  users
• Relevance to research at J9 and potential to
  integrate our efforts
• Our initial CPoF work
• CPoF data extraction analysis
• Related survey data collection analysis
• Summary of results
• Potential to leverage in a near-term
  collaborative effort with J9
• Our C2insight Phase II work
• The development plan for C2insight
• Integration with Urban Resolve 2015 in further
  work with J9

35

• An initial CPoF MetaMatrix
• Entities are identified uforms (or aggregations
  of uforms) of particular interest
• Relationships are instantiated by explicit uform
  links or calculated relationships (e.g. proximity
• Metrics are network graph metrics
• Node level
• Network level
• Organization level

traditional social network analysis
36
Straightforward and achievable
More difficult(tractable given some assumptions)
37
Objectives for for C2insight

• Utilize CPoF generated data
• Utilize the power of network analysis for finding
  significant patterns
• Provide contextual information to help users
  explore interpret results
• E.g. geographic, temporal, significant events
• Make the system usable by applying human factors,
  e.g.
• Overview display principles overview, relate,
  filter, details-on-demand, zoom, maintain global
  perspective, layering separation
• Support user/role specific reports
• Support configuration adaptation
• Develop iteratively involving SMEs in design
  evaluation
• Field test

In Phase II the goal is to build a tool that the
command staff can use themselves
38
Information in Context
Again a major goal is information in context
39
C2insight as a CPoF appliance could leverage the
collaborative environment and that interface to
CPoF data
40
Histogram of activity can be used to select
period of interest
C2insight filters are applied using CPoF Look
and Feel
Social network relationships are displayed along
with type and geographical information on CPoF map
Selected tasks appear highlighted across
different visualizations and reports
Selected tasks appear highlighted across
different visualizations and reports
Selected tasks appear highlighted across
different visualizations and reports
Concept of C2insight components integrated with
the CPoF desktop
41
Phase II C2insight Schedule
Core software development will be interleaved
with prototype development and SME feedback
42
C2insight Milestones

• October 1, 2006
• Search interface mock-up
• October 15, 2006
• Communication activity front-end prototype
• November 1, 2006
• C2insight baseline infrastructure implemented
• Geography-Social Network front-end prototype
• November 20, 2006
• Semi-annual technical report
• December 1, 2006
• C2insight requirements specification
• ORA integration into C2insight
• March 15, 2007
• C2insight able to read, analyze, and visualize
  CPoF data
• Revised requirements document
• April 1, 2007
• Automap integration into C2insight

43
C2insights Architecture
44
C2insights Architecture
Multiple backend data sources (CPoF and other)
may be connected, appearing as one through the
data interface
Set of analysis modules allows a variety of
algorithms, including SNA, to be applied to the
data of interest
Data model serves as virtual database, ensuring
all components are independent of the actual data
sources
A framework for providing and combining multiple
linked views of the information
UI allows one to select data of interest, desired
analyses, and visual form for presenting the
information
45
Vision for integration with Urban Resolve 2015

• Phase II
• Refinements based on Phase I feedback
• Near real-time results (hours)
• Analysis of organizational effects resulting from
  C2insight use
• Phase III
• Refinements based on Phase II feedback
• Results in real time
• Analysis of effects resulting from real-time
  C2insight use

46
BACKUP SLIDES FOLLOW
47
C2insight Organization
Small group with roughly equal mix of human
factors, social network, and software experts
48
Net-Centric Operations in Action

• When DARPA contractors teamed with Soldiers from
  the 1st Cavalry Division to take this
  experimental technology to war, it was a huge
  leap of faith on the part of a visionary leader
  and a dedicated group of his Soldiers.
• (Maj Ryan Paterson, USMC. DARPA CPOF PM.
  DARPATech 2005).
• I think it is going to have an impact, not only
  on tactics, techniques, and procedures we use to
  command and control. I think it is going to have
  an impact on the size of our staffs, and what our
  staffs do. The wargaming process is so critical
  to the Military Decision Making Process, it
  CPOF is going to change everything we do and
  how we fight. I, in 32 years in the Army, have
  never seen a single system that will have a
  greater impact on our Army and our entire Joint
  Force than CPOF.
• (MG Peter Chiarelli, 1st Cavalry Division, US
  Army)

49
CPOF Vision
50
Network Centric Warfare Conceptual Framework
Alberts, D. S. and Hayes, R. E. (2003). Power to
the Edge Command and Control in the Information
Age. CCRP (Jun 03) 18.
51
C2insight - Our Multidisciplinary Approach

• Dynamic Network Modeling Analysis
• What relationships between people can be
  analyzed?
• What other entity-relationships might be useful?
• What network graph metrics are applicable here?
• CPOF Data Structure Analysis
• What is in the repository and how is it
  represented?
• What aggregation would be useful?
• Cognitive Task Analysis
• What can we learn from observations, interviews,
  and participatory design sessions?
• What are the cognitive processes involved in
  using the system as it is?
• What is the workflow?
• Visualization Needs Analysis
• How can we make important relationships visible
  at a glance?
• How should we support user adaptation of views on
  the system?
• How should we integrate analysis and exploration?

52
Dynamic Network Modeling Analysis
53
Event Initiated by, modified by, modified by,
Analysis of Text Content
54
VOIP Feedback Analysis
select relationmemberOf, talkingTo, precedes, .
.
select directmanipulation behavior
select time
55
Three scenarios developed with General ONeal

• The Crisis Intelligence reveals insurgents are
  planning to kidnap a government official in the
  Green Zone.
• The ADC-M wants to know what activity has
  occurred in the area and who has knowledge about
  it
• The results can impact who is brought into a
  planning session
• The patterns of activity are used to make the
  planning session more informed and can help
  decide where to place trusted agents of a
  supportive religious leader
• Transfer of Authority The Liberty Brigade is
  assuming command and control from the Mountain
  Brigade for a particular AOR, and there is
  concern the insurgents will seek to benefit from
  the new Brigades and the Divisions need to
  adapt
• The ADC-M wants to compare these organizations
  what are their respective patterns of activity?
  How are the brigade staff collaborating with the
  division?
• As he listens to their analysis he can utilize
  his own independent assessment of what has
  occurred and his knowledge of what they have been
  doing.
• The Corrupt Chief of Police A particular Chief
  of Police, suspected of being corrupt, controls
  10 local police stations.
• Are there more or less incidents in these regions
  compared to other comparable regions?
• Who has knowledge about these incidents?
• What type of data has been collected and what do
  we know about this Chief of Police in particular?

56
Example applications of these kinds of patterns

• Briefings
• Pre-BUB and CUB automated reports for CG and
  others
• Provides the CG with an independent assessment
• Facilitates better use of time
• Presentations during briefings
• What are the red force patterns?
• What actions seem to reduce attacks on us?
• During briefings question answering
• Intelligence Analysis
• Is there evidence to support a tip that a police
  chief in one area is corrupt?
• Planning
• Deciding who to bring into a planning session
  based on their knowledge
• Execution
• The Division battle captain tracking expected
  activities of the Brigade CPoF users
• General
• Annotation of CPoF icons based on risk factors

57
Information Visualization Requirements
58
Feedback for Adaptation Interactive Networks in
Context

• More informed and efficient briefings,
  intelligence analysis, planning, execution,
  re-planning. e.g.
• Pre-BUB CUB independent reports for CG and
  others
• Real-time discussion of patterns during
  briefings or planning
• Real-time tracking expected activity
• Planning-team member selection
• Evidence collection for hypothesis testing

red blue force patterns of behavior over time
workstation
event icon
abstract views coupled with geographic maps
provide insight
interactive legend
geosticky icon
temporal context
proximity relationship
geographic context
filters
created by relationship
familiar frames of reference with direct
manipulation
Boundary Spanners (high betweenness, low
degree) 1 CG4ID (1759.826) 2 TAC24ID
(1175.842) 3 PLANS24ID (762.8000) 4
Helpdesk (451.6840) 5 CICStrike (299.9726)
Organizational Structure WORKSTATION count 36
Overall Complexity 6.0949e-002(Density of the
Meta-Matrix ) Overall Components 6(Number of
Meta-Matrix components)
user/role specific report
59
Organizational Relationships in Context Under
User Control
Entities Relationships
ER - key
Annotated Temporal Overview Zoom Control
Geographic Overview Zoom Control
Specific Geographic Context
Animation Control
Entity Relationship Filters
Temporal Filters
60
(No Transcript)