Key Sensor Concepts

From Semantic Sensor Network Incubator Group
Jump to: navigation, search

This page describes the key concepts in describing sensors. It sets out the main concepts in describing sensors and the categories into which those concepts fit.


The discussion mainly relates to the two figures on this page. The first figure is based on a review of concepts in existing ontologies for sensors (see the group's review and the draft of the survey). It describes concepts that are present across a number of these ontologies and how the concepts for describing sensors can be broadly split into categories. The second figure is a basic outline of the CSIRO ontology and how its versions of the key sensor concepts are organised.



Diagram 1 - basic sensor concepts

The concepts in the various sensor ontologies can be largely split into five groups: those describing sensors and systems, those describing the capabilities of a sensor, those describing processes, those describing physical aspects of sensors, and concepts for observations. (Note: this page is mainly about sensor concepts, a similar outline for observations is yet to be done)


Highlevel.gif


Depending on the particular focus of the individual ontologies, more or less detail is supplied for any given set of concepts. All the ontologies have a concept for sensor - though with different intended interpretations - and a number have concepts for systems, components etc, and how these are organised. Most of the ontologies have concepts describing accuracy, etc, though only the CSIRO ontology seems capable, for example, of describing different capabilities in different conditions (for example, accuracy that depends on conditions). Some form of process can be described in many of the ontologies, again with varying degrees of expressive power - ranging from part-of relations to structural composition. There are a range of physical properties that can be described from location and platform in most to operating restrictions in MMI.


Diagram 2 : how the basic concepts are organised in the CSIRO ontology

This diagram takes the key concepts for sensors (those identified across a number of ontologies as above) and shows how they are related in the CSIRO ontology. The dashed lines indicate concepts that are in a number of other ontologies but not yet in the CSIRO ontology (except PredictiveModel which is included to illustrate that the sensor concept is intended to capture a range of sensing devices).

KeyConcepts.gif

A sensor may have a number of functions or operations each of which is modelled by an OperationModel (which may be modelled black-box or white-box including the process internals). An OperationModel can specify the properties of the operation in any number of conditions, including for example different accuracies in different conditions and specifying the range of measurement of the sensor. The process modelling is intended to model both computational processes and physical processes (input and output can refer to any sort of parameter - a factory takes mud as input, the process heats them and it outputs bricks).

The ontology is intended to model sensor types (temperature sensor, wind speed sensor, etc), commercial sensors (a manufacturer's wind sensor, for example), and individual sensors (particular sensors in a deployment, for example). Sensor types and manufacturer's sensors are largely modelled in the DL TBox, while individual sensors are modelled in the ABox. For example, a company X wind speed sensor is modelled as TBox concepts specifying the capabilities and physical properties of the sensor from its data sheet; instances of such a sensor are then modelled as ABox individuals (though much of the specification is implied by the TBox concepts and needn't be repeated - location, orientation etc would be specific to individual sensors though).