Print This Post

GeoWorld: Breaking the Barriers – GML and WFS Enhance Sensor Data

GeoWorld August 1, 2006

[Get Copyright Permissions]
Copyright 2006 M2MEDIA360

By Ron Lake and Frank DeNap

Ron Lake is chairman and CEO, Galdos Systems Inc.; e-mail: rlakeatgaldosdotcom. Frank DeNap is SensorNet project manager, Oak Ridge National Laboratory; e-mail: denapfaatornldotgov.

Oak Ridge National Laboratory (ORNL), a leading U.S. Department of Energy organization, has been developing wide-area sensor networks in the SensorNet project for a variety of purposes, including applications to enhance military installation security as well as transportation security at U.S. ports and highway systems. Such systems must combine information from legacy and newly developed measurement systems as well as provide integrated information to a variety of distributed stakeholders.

Furthermore, information needs to be sourced from numerous jurisdictions and networks with different security constraints and distributed across multiple agencies. In this era of heightened threats and instant communications, it's critical that governments be able to share geographic sensor information seamlessly, accurately and systematically-regardless of jurisdiction or role.
For these reasons, ORNL has been a leader in the investigation and application of open standards in the development and deployment of spatio-temporally aware wide-area sensor networks. A vital component is the Open Geospatial Consortium (OGC) Web Feature Service (WFS) and Geography Markup Language (GML).

A diagram highlights WFS locations in a wide-area sensor network architecture.

WFS and GML are central components of many spatial data infrastructures, because they enable vendor- and jurisdiction-neutral sharing of geospatial information in near real time. Although many people may think that this is restricted to geographic features such as rivers or roads, WFS and GML are equally applicable to wide-area sensor networks involving near-real-time sensor data.

Sensor Data

Collecting, managing and distributing sensor data is complex. Sensor devices vary a lot in terms of the quantities measured, units of measurement and the manner in which the measured data are encoded.

In addition, security applications require that such data be associated with geographic features and integrated with other sensor data. This is essential to enable the rapid use of data in decision-making activity.

A truck weigh station is a practical example of a deployed wide-area sensor network, measuring vehicle and container weights and density as well as passing the information along a highway network.

Measured quantities of interest in security applications are wide ranging and can include chemical-species concentrations, radioactivity, vehicle and container weights, and density as well as common meteorological parameters.

Representing Data with GML Observations

Using GML technology, sensor data can be acquired and managed through the creation of observations. A GML observation models the act (or the doing) of the observing-it's a GML feature.

GML observations have the following properties:

  • location (point, line, area) of the observing
  • validTime of the observing
  • resultOf the observing (this can be a link)
  • using—how the observing was done (what was used in doing it, which can be a link)
  • subject or target of the observing (also can be a link)

There's also a concrete DirectedObservation, which adds a "direction of looking" to the aforementioned properties. (For more information, see www.geoplace.com/uploads/FeatureArticle/0505gml.asp.)

In most cases, a user would develop a domain-specific application schema based on GML observations, and this is the case for the ORNL SensorNet projects. Within ORNL's system, some observations may reflect the existence or characteristics of geographic features (e.g., change in vehicle position or path), while others reflect radioactivity and other types of sensor measurements.

A weather-sensor station is one of many possible arrays in a wide-area sensor network.

Observations can be created by sensors or humans and transmitted in the SensorNet as WFS transactions or responses. WFS transactions could, for example, be emitted directly by a sensor or through a proxy or gateway. Such transactions enable the WFS to behave as an "intelligent data repository."

GML and WFS in SensorNet

The use of WFS in wide-area sensor webs is a natural extension of GML, which can express a wide range of geographic information types, including conventional geographic features as well as coverages and observations. The ability to handle such data is a strength of WFS, because it minimizes the number of different interfaces and components in the sensor network.

In addition, the sensors themselves may be moving or located on moving platforms. And they can be used to determine and interpret the values of conventional feature properties as well as be used in the context of such features.

ORNL selected WFS as a primary vehicle for the distribution of, and access to, real-time sensor data, because of its ability to deal in a generalized manner with sensor data and other types of geographic features. WFS also was chosen because several commercial and open-source implementations currently exist.

Other sensor-web technologies, such as the OGC Sensor Observation Service (SOS), are emerging from the standards process. Although the newer standards have promise, they don't yet have a variety of available implementations. In addition, it should be noted that the domain of applicability of SOS is sensor observations and not feature data in general.

Commercial implementations of transactional WFS include products by Galdos Systems, Ionic Software, CubeWerx and Intergraph. In addition, open-source components and tools such as GeoServer and the UMN Map Server provide varying degrees of WFS support. ORNL currently is using the Cartalinea WFS from Galdos Systems in some of its deployments.

Commercial WFS products such as Cartalinea provide several features that were important to ORNL, including support for GML 3 (e.g., GML observations) and the ability to invoke external actions (e.g., send notification messages) based on changes in data. Such external actions can include the notification of authorities as well as the replication of features to other data storage systems in SensorNet.

An important issue for the deployment of web services in SensorNet is that of performance and scalability. The generality, flexibility and extensibility of web services need to be balanced against the ability to handle required sensor-update rates from a reasonable number of sensors. ORNL has been working with several vendors on these issues.

Radiological Schemas

Every WFS represents a collection of objects described by one or more GML application schema. In the case of SensorNet, the objects are measurements of radiation data that are encoded as GML observations.

Radiological measurements are an important aspect of SensorNet deployments. To standardize radiological measurements, ORNL adopted the ANSI N42.42 data standard, which provides standard names for various sensor components and readings. It also can be used with a variety of radiation detectors, including simple spectrometers, radiation pagers, radionuclide identifiers, PRD/survey meters, gross counting portal monitors and spectroscopic portal monitors.

To make use of this schema with a WFS, ORNL recasts it as a GML application schema using GML observations. This schema will be submitted by ORNL and Galdos Systems to OGC for consideration as a "best practice" in the application of GML to sensor data systems. The schema preserves the standardized names and usage from ANSI N42.42, but it uses GML structures and primitives for geographic location and the measurement of time.

From Pilots to Full Deployment

ORNL's SensorNet projects are currently in the pilot stage, investigating issues of feasibility, scalability and performance. The pilots are aimed to develop standardized system architectures that can be deployed in operational systems in domains such as force protection and military installations as well as transportation security for U.S. ports and highways.

ORNL's standardization approach enables commercial vendors to start serving the requirements of large-scale, robust, sensor-network deployments for government and private-sector customers. Operational versions of these systems will be deployed during the next two years to help with the automated inspection and tracking of containers from port of entry to final destination within the United States.

Using a transactional WFS to gather and share sensor data, users can create value-added applications based on the collection, integration and distribution of information from diverse wide-area sensor networks across multiple jurisdictions.

This article was originally published in GeoWorld magazine. Download a PDF copy: Breaking Barriers-GML and WFS Enhance Sensor Data.