Introduction Introduction
ToTo
SensorMLSensorML
Alexandre Robin – October 2006
SensorML – Design Objectives
Standard way of describing wide range of sensors and sensor systems (platforms, sensor grids…) Electronic Datasheet
Enable sensor discovery among high number of disparate sensors accessible through a network
Integrate within OGC Sensor Web Enablement Framework
Alexandre Robin - October 2006
Allow precise description of complex systems
Keep simple case simple! (i.e. thermometer)
Allow global cross-domain classification
Allow local/specialized domain specific classification
SensorML – Design Objectives
Alexandre Robin - October 2006
Describe precise lineage of data, with enough information to allow error propagation
Facilitate data processing and geo-location Automatic
Provide enough information to understand and simulate sensor behavior
With small human intervention in the general case
Automatic processing within a specific domain/profile
SensorML – What can be described?
Alexandre Robin - October 2006
Platforms and Constellations
SML SystemSensors & Models
SML SystemSML Component
Raw Data
• Nature• Structure• Encoding
Data Processing
SML ProcessModelSML ProcessChain
Data Product
• Nature• Structure• Encoding
SensorML – What can be described?
Alexandre Robin - October 2006
Frequency Response
GeometryCharacteristics
- WHAT was measured? Phenomenology, Frequency Response
- HOW was it measured? Calibration, Quality
- WHERE was it measured? Geometry, Spatial Response & Sampling
- WHEN was it measured? Temporal Sampling, Impulse Response
- WHY was it measured? Application
SensorML – Sensor Systems
Component 1Thermometer
System – Weather Station
Component 2Barometer
Component 3Anemometer
AirTemperature
AtmosphericPressure
WindSpeed
Digital Number
Component 4Processing
Digital Number
Digital Number
WindChill
Temp
Alexandre Robin - October 2006
SensorML – Sensor Systems
Alexandre Robin - October 2006
System – Aircraft Platform
GroundRadation
AircraftPosition
Subsystem 1: Scanner
DetectorBand 1
DetectorBand 3
DetectorBand 2
DetectorBand 4
Subsystem 2: INS
GPS
IMU
InterleavedScanline
GPS Data Tuple
IMU Data Tuple
SensorML – Header Info
Alexandre Robin - October 2006
Keywords, Identifiers and Classifiers for classification and indexing in Registries and Catalogs
Global Characteristics and Capabilities for quick view on System capabilities
Relevant Contacts and Documents to point to additional knowledge and documentation
Temporal, Legal and Security Constraints to make sure the document is used only when appropriate
History to keep track of System changes such as calibration events or other modifications
SensorML – Inputs, Outputs, Connections
Alexandre Robin - October 2006
Specify nature of measured phenomena. Points to dictionaries which provides robust cross-domain semantic associations
Specify units of measure for each scalar component of the inputs and outputs
Specify quality of values and constraints (interval, enumeration)
Possibilities of grouping and defining arrays of values as input and output
Define connections between components to describe their interactions within a System
SensorML – Relative Positions
Platform
GPSIMUScanner
Swath
Alexandre Robin - October 2006
Relative positions ofSystem components
(Both location and orientation!)
Reference Frames ofSystem components
(How it relates to hardware)
SensorML – Detector Component
IdentifiersClassifiersConstraints
Detector
ContactsDocumentationReferences
CharacteristicsCapabilities
GeometryTiming
Spatial FrameTemporal Frame
Response Characteristics
Additional information used for detail discovery and
link to other documents
Sensor internal geometry (look rays
direction for a scanner or camera)
Definition of coordinate frames
attached to the sensor
Identification and Classification terms for further discovery
Sensor timing (look rays times for a scanner = gives time sequence)
Response characteristics
(calibration, error, frequency)
Alexandre Robin - October 2006
Inputs OutputsP
aram
s
SensorML – Detector Response
Alexandre Robin - October 2006
Calibration
in
out
Random Error
%
q
Spectral Response
dB
Impulse Response
t
dB
Spatial Response
Temporal Response
t
dB
Integrationtime
Calibration CurveGives the mapping of input to output values for a steady state regime. Two curves are used to describe a Hysteretic behavior.
Random Error CurveGives the relative measurement error versus the input value itself or any other environmental quantity such as temperature.
Spectral Response CurveSpecifies dynamic characteristics of the detector in the frequency domain. It gives the sensitivity of the detector versus the frequency or wavelength of the input signal.
Impulse Response CurveSpecifies dynamic characteristics of the detector in the time domain. It represents the normalized output of the detector for an impulse (D function) input.
Spatial Response Curve(s)Gives the sensitivity of the detector relative to spatial coordinates (location of the source, or orientation of the incoming signal, e.g., point spread function, polarization)
Temporal Response CurveGives the sensitivity of the detector relative to a temporal coordinate frame (e.g., sampling time). This is a more descriptive form of the integration time.
SensorML – Component Array
Alexandre Robin - October 2006
Concept of SensorML Array can be used to describe arrays of any Component or System
Powerful to describe large arrays of “almost” identical devices
Ability to individually tweak elements of the array through an indexing mechanism
SensorML – Detector Array
Alexandre Robin - October 2006
Sensor (CCD array)
Detector R1
R0
Radiance
Cell Index
DN [3000]
DN (16 bits)
Look Up Table
Table Data
Calibration
Spatial Sensitivity
Gain
Corrective Gain
Position (R1 vs. R0) (Sampling)
X Translation
Y Translation
Z Translation
X Rotation
Y Rotation
Z Rotation
Index
10º
-10º
Z
X
Y
R0
SensorML – Processing Chain
IMU and GPS sensor data
Look Up Table
Look Up Table
ScanIndex
T T
TimeInterpolator
ScanTime +
Look Ray Time
Look Ray Position
Adjusted Time
INSData
LLAPoint
LLA To ECEF
IFOIGeometry
EllipsoidIntersectionPosition in
sensor CRSPosition inECEF CRS
Position of INS in LLA
Position of INS in ECEF
Derived from relative positions of
sensors
Obtained from Sensor Geometry
(FOV…)
Obtained from Sensor Geometry
and Timing
Alexandre Robin - October 2006
SensorML – Data Description
Alexandre Robin - October 2006
Scanline
Time DataArray … (x 720)Radiance Radiance Radiance
Specify Data Structure (imagery, in-situ, spectral, …)
Weather Data
Time Temperature PressureWind Speed
Spectrum
Time DataArray … (x 250)Freq1 Freq2 Freq3
SensorML – Data Description
Alexandre Robin - October 2006
Specify Data Structure (imagery)
Image RGB (1024x768)
DataArray
… (x 768)
DataArray … (x 1024)DataGroup G BR
DataArray … (x 1024)DataGroup G BR
SensorML – Data Description
Alexandre Robin - October 2006
Specify Data Encoding (ASCII, Base64 binary, Raw binary)
Data structure can be described in the interface section of a System/Component
Specify parameters for each scalar value in the structure
Can specify compression methods and encryption
Data structure can be described separately along with the observation values
Relevant Links
Open Geospatial Consortium
http://www.opengeospatial.org
SensorML
http://vast.uah.edu/SensorML
Questions?
Alexandre Robin - October 2006