Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out
Your Federal Quarterly Tax Payments are due April 15th Get Help Now >>

SRI _ ESD

VIEWS: 8 PAGES: 27

									 ONISTT Spatial Ontologies
              June 20 2006

              David Martin
            SRI International

Work done with Reg Ford, Jonathan Solnit,
Mark Johnson, Daniel Elenius, Dave Hanz
ONISTT Overview


 Open Net-centric Interoperability Standards for
 Training and Testing (ONISTT) is:
  • A project sponsored by Deputy Under Secretary of Defense
     for Readiness (DUSD/R)
   • Researching means to automate the composition of
     improvisational confederations of heterogeneous systems to
     support specific mission enactments
 A fundamental functionality required within such a
 confederation is the sharing of time-space position
 information (“TSPI”)
   Need way to identify all possible ways a system could represent
    TSPI in a form that can be understood by an Inference Engine
      Need ontology for spatial concepts
Framing the Problem



  Executing complex operations requires more capabilities
     than are resident within any single system
    Need capability to establish a assemblage of
     heterogeneous systems to perform a larger task:
       • a System-of-Systems (SoS)
  This requires achieving interoperability among the
     constituent systems in the SoS:

     Interoperability: “The ability of systems, units, or forces to
     provide services to and accept services from other systems,
     units, or forces and to use the services so exchanged to enable
     them to operate effectively together.”
                                      (definition from Joint Pub 1-02)
          Assembling A System of Systems
          Deliberately Planned versus Improvisational

   The traditional approach for assembling a SoS has relied on deliberately
    planned assemblage of cooperating systems, where
      •   All possible member systems and their roles in the SoS, and
      •   All required functional capabilities of the SoS, and
      •   All Information Exchange Requirements, data formats, etc.
      … are identified & rigorously defined long before (typically several years before) the
       actual operation commences
      •   Plus: there is some administrative body in charge of assembling the SoS which has
          sufficient authority to mandate changes in some (or all) of the constituent systems to
          achieve the goals of the SoS
   Recent history of Joint and Multi-national operations have shown that the
    deliberately planned SoS approach does not provide adequate flexibility to
    support the kinds of operational capabilities that are now needed
       Need a new approach that will support formation of an Improvisational assemblage
          of heterogeneous systems on short notice – effectively creating a SoS – but
          without the SoS “baggage” listed above
            ONISTT – How it works
Capture
domain
knowledge                                             Training
                                                      System




             Domain




                                    Web Services
            Knowledge




                                     Interface
                                                       Training
                                                       System




                                                   Training
                                                   System
           ONISTT – How it works
Create
Metadata
for                                                  Training
                                                     System
Systems
                       Systems
                       Registry


            Domain




                                   Web Services
           Knowledge




                                    Interface
                                                      Training
                                                      System




                                                  Training
                                                  System
               ONISTT – How it works
Specify
Scenario
Requirements                                             Training
                                                         System

                           Systems
                           Registry

 Scenario
 Spec           Domain




                                       Web Services
               Knowledge




                                        Interface
                                                          Training
                                                          System




                                                      Training
                                                      System
                 ONISTT – How it works
Generate
candidate
confederations                                             Training
                                                           System

                             Systems
                             Registry

 Scenario
 Spec             Domain




                                         Web Services
                 Knowledge




                                          Interface
                                                            Training
                                                            System



                  Analyzer


                               Config-
                               uration                  Training
                 Role –        Spec
                 Actor                                  System
                 Mapping
              ONISTT – How it works
Configure
systems for
exercise                                                        Training
                                                                System

                             Systems
                             Registry

 Scenario
 Spec          Domain




                                              Web Services
              Knowledge




                                               Interface
                                                                 Training
                                                                 System
                                  Mediators


                Analyzer


              Configurator     Config-
                               uration                       Training
                               Spec
                                                             System
     Augmented map projections

     Augmenting map coordinates
        with ellipsoidal height produces a 3D CS.
 MP coordinate (u, v)                             (, ) surface geodetic coordinate
Augmented
MP coordinate (u, v, h)                           (, , h) 3D geodetic coordinate

              (u 1 , v1 , w)                          (u 2 , v2 , w)



             (u 1 , v1 , 0)    coordinate-space           (u 2 , v2 , 0)




                    p’                                         q’


                         p                            q
                               position-space
  Augmented map projections


Line of sight visibility in
augmented map projection coordinate space...
        (u1 , v1 , w)                       (u2 , v2 , w)



       (u1 , v1 , 0)    coordinate -space       (u2 , v2 , 0)




              p’                                    q’


                   p    position-space
                                            q

      … can be an illusion.
ONISTT Selection of Reference Spatial Standard


   ISO-19111: Geographic information - Spatial referencing by
    coordinates
     •   Used primarily for GIS applications
     •   Existing ontologies


   ISO-18026: Information Technology Spatial Reference Model
     •   Based on SEDRIS standards
     •   Common reference model used for modeling and simulation (M&S) and
         Live-Virtual-Constructive (LVC) training system applications
     •   Specifies transforms
     •   No existing ontology
          •   But structure mapped naturally into an ontology framework (OWL)

   A National Geospatial-Intelligence Agency (NGA) report compares
    19111 and 18026
     •   Concludes that the scope of 18026 is much broader than 19111 and that
         19111 is a subset of 18026
SEDRIS SRM


   The SEDRIS Spatial Reference Model (SRM) provides a precise
    specification of geometric properties such as position (location),
    direction, and distance

   SRM is one of five parts of SEDRIS
     •   Data Representation Model (DRM)
     •   Environmental Data Coding Specification (EDCS)
     •   Interface Specification (API)
     •   SEDRIS Transmittal Format (STF)


   SRM is codified in ISO:18026
SRM Composition


   Goal is to specify a spatial coordinate system in which positions can
    be identified

   SRM accomplishes this through a spatial reference frame (SRF)
     •   Combines an abstract coordinate system (ACS) and an object reference
         model (ORM)




             Figure taken from ISO-18026
ACS and ORM


   An ACS associates coordinates with positions in an abstract
    Euclidean space, called its position-space1
     •   Examples include Euclidean 3D (x-y-z) and Spherical


   An ORM defines a precise relationship between a position-space and
    an object-space1
     •   Object-space is specific to the object of interest (e.g. the Earth)
     •   Includes normal embedding, a length-preserving function between
         position-space and object-space
     •   Example is the well known WGS 1984




              1 – Quoted from ISO-18026
Reference Datum


   A Reference Datum (RD) is a geometric primitive that relates
    measurements and/or geometric characteristics of object-space to
    position-space1
     •   Examples are spheres or ellipsoids


   ORMs use RDs to specify the position-space to object-space
    relationship, and to model spatial objects
     •   An ellipsoid typically is the model for the Earth




             1 – Quoted from ISO-18026
Building an SRM Ontology


   Separated each main idea into
                                         Import Tree
    individual ontologies
     • ACS
                                           SRF
     • RD
     • ORM
     • SRF
                                    ORM          ACS




                                    RD
Subclasses


    Each of the four ontologies have subclasses defined in 18026, mostly
     based on number of dimensions
    RD subclass “Physical” is for specific objects such as oblate
     ellipsoid




    Ontology         Subclasses
    ACS              1D, 2D, 3D, surface, map projection
    RD               2D, 3D, and “physical”

    ORM              2D and 3D
    SRF              2D, 3D, surface, map projection
Properties


   Properties came from tables in 18026 describing the characteristics
    of each class
   Created enumerated classes for all properties that had a known set of
    values



Ontology             Properties
ACS                  Code, properties (orthogonal, orthonormal, etc.), descriptor
                     (3D linear, surface map projection, etc.), function type
                     (generator or map projection)
RD                   Code

ORM                  Code, reference datum set
SRF                  Code, ACS, ORM, object type (physical, Earth, satellite, etc.)
Individuals


   ISO:18026 specifies instances of each of the concepts

   Currently our ontology only has those instances necessary to specify
    Geocentric WGS1984 and Geodetic WGS1984

   Also have individuals for enumerated classes
  Example: Geocentric WGS1984

srf:GeodeticWGS1984
           srf:absCoordSys
                             acs:Geodetic
                                   acs:csProps
                                                    acs:orthogonal
                                   acs:descriptor
                                                    acs:ThreeDCurvilinear
                                   acs:fcnType
                                                    acs:generatingFunction
           srf:objRefMod
                             orm:WGS_1984
                                orm:refDatumSet
                                                    rd: origin3D
 Purple text: individual
 Blue text: property                                rd: zAxis3D

                                                    rd: zPlane3D
Future Work


   The current ontology is not a complete specification of ISO:18026

   Possible additions include:
     •   Coordinate system transformations
          •   Includes identifying the properties that are preserved or not when transforming
              from one coordinate system to another

     •   Temporal coordinate systems
     •   Vertical offset surfaces
     •   “Generating functions” for abstract coordinate systems
     •   Binding constraints for ORMs
     •   Filling in the rest of the individuals
Other ONISTT Spatial Ontologies


   Because few standard domain ontologies currently exist, the ONISTT
    team developed some preliminary (temporary) custom ontologies for
    small subsets of fundamental domains
   Spatial ontologies
     •   ISO-18026 ontology provides spatial reference frame property for entity
         positions
     •   Positional component properties: EngineeringValue and
         EngineeringMeasurement
     •   Time-Space Position Information (TSPI) ontology adds time to positional
         information
          •   Time ontology extends daml/time-entry

   Example ONISTT application of spatial ontologies
     •   Describe inputs, outputs, quality characteristics for service ontologies
     •   nexri/tspi-report ontology completely describes the contents of a
         message, from semantics to bit representation
ONISTT Time Space Position Information (TSPI)
               Import Tree
ONISTT Engineering Value Ontology


   EngineeringValue class
     •   Properties: magnitude and unitOfMeasure (EngineeringUnit)
     •   Subclasses for Length, Angle, Quaternion, etc.


   EngineeringUnit class
     •   Subclasses for LengthUnit, AngleUnit, etc.
     •   Property siUnit identifies the standard unit for each unit class
     •   Property siUnitConversionFactor identifies the conversion from each
         subclass individual to the standard unit
          •   E.g., for the AreaUnit acre individual, the siUnitConversionFactor to the siUnit
              (square-meter) is 4.046873E+03
ONISTT Engineering Measurement Ontology


   Adds concept of measurement uncertainty to Engineering Value
   EngineeringMeasurement class
     •   Subclasses for LengthMeasurement, AngleMeasurement, etc
     •   Property measurand is an EngineeringValue
          •   E.g., the measurand value for LengthMeasure subclass is restricted to the
              Length class of EngineeringValue

     •   Property uncertaintyType is Uncertainty class
     •   Property uncertaintyValue is EngineeringValue type (restricted as
         appropriate for the subclass) or FractionalUncertainty type
   Uncertainty class
     •   Property estimateType is UncertaintyEstimateType class, with individuals
         CEP, SEP, etc.
     •   Property distributionType is UncertaintyDistributionType class, with
         individuals Gaussian, Poisson, Rayleigh, etc.
ONISTT Time Ontology


   ONISTT extended damltime/time-entry.owl (by Feng Pan and Jerry
    Hobbs) to provide to M&S timestamps, adding the following classes:
     •   Granularity – smallest distance between two instants supported
     •   Reference – starting point of a time standard, e.g. midnight January 1,
         1970
     •   Rollover – when a time resets to 0 (to avoid overflow of a limited size time
         field)
     •   Timefield – class with object properties to use the first three, plus
         datatype properties for the timefield size in bits and time value
          •   Time value is a TimeInstant EngineeringMeasurement (including units and
              uncertainty estimate)

     •   Timestamp – class with an object property specifying one or more
         Timefields
   Allows unambiguous specification of the various M&S time formats
    encountered by the ONISTT program

								
To top