A Semantic Sensor Web for Environmental Decision Support Applications

Alasdair J G Gray,Kevin Page,Asunción Gómez-Pérez,Jean-Paul Calbimonte,Alvaro A A Fernandes,Ixent Galpin,Oles Kit,Jason Sadler,Kirk Martinez,David De Roure,Alex Frazer,Manos Karpathiotakis,Raúl García-Castro,Norman W Paton,Manolis Koubarakis,Oscar Corcho,Kostis Kyzirakos

doi:10.3390/s110908855

Abstract

Sensing devices are increasingly being deployed to monitor the physical world around us. One class of application for which sensor data is pertinent is environmental decision support systems, e.g., flood emergency response. For these applications, the sensor readings need to be put in context by integrating them with other sources of data about the surrounding environment. Traditional systems for predicting and detecting floods rely on methods that need significant human resources. In this paper we describe a semantic sensor web architecture for integrating multiple heterogeneous datasets, including live and historic sensor data, databases, and map layers. The architecture provides mechanisms for discovering datasets, defining integrated views over them, continuously receiving data in real-time, and visualising on screen and interacting with the data. Our approach makes extensive use of web service standards for querying and accessing data, and semantic technologies to discover and integrate datasets. We demonstrate the use of our semantic sensor web architecture in the context of a flood response planning web application that uses data from sensor networks monitoring the sea-state around the coast of England.

Highlights

Sensor networks are increasingly deployed to monitor the state of the physical environment around us
We have presented a semantic sensor web architecture that comprises a core collection of services that form a service-oriented architecture for data publication, discovery, and integration
The semantic sensor web architecture supports the four capabilities identified in Section 1: (1) identify relevant sources of data; (2) access sensor data in near real-time; (3) correlate disparate heterogeneous sources; and (4) support multiple conceptualisations

Summary

Introduction

Sensor networks are increasingly deployed to monitor the state of the physical environment around us. To effectively predict when a surge is going to take place requires gathering data from a wide variety of sources published by independent autonomous providers: sensor networks that monitor the status of the sea provided by research institutions, government agencies, and private companies; weather forecasts provided by national meteorological offices, and companies; and coastal defence information provided by government departments. These are used as inputs to environmental models which predict the future sea-state, and the probabilities that sea defences will be breached or over-topped. Planning the response to a potential flooding event requires a large number of additional sources to be available (e.g., shipping, traffic, and man-made assets), which can be related to the results of the forecast and the current conditions

Objectives

Discussion

Conclusion