Archner, O*:
Environmental Sensor Observations
Vortrag (eingeladen),
ETNA School on Plant Phenotyping, Jülich: 01.11.2009 - 10.11.2009
Abstract:The current hardware and software developments in the field of wireless sensor networks and embedded devices have stimulated a tremendous change in the environmental sensor observation domain. Not that long ago data were logged by using analogue chart records. Today rather small independent devices called wireless sensor motes[4] have taken their place. These devices are able to measure analogue signals, convert them into digital signals, and transmit the values via wireless networks. Due to their rather cheap price and their excellent functionality their propagation is inevitable. The so called ‘Sensor Web’ is knocking at our doors[6; 3]. How about the application of these techniques? What are the key issues to be aware of, if we would like to build a reliable sensor network to observe e. g. micrometeorological parameters[2]? Is it really as simple as proposed by the various manufacture announcements? What kind of pitfalls exist when we plug our micrometeorological device into a wireless sensor network? In his talk the author tries to answer these questions. Starting with the typical use case of measuring the green house air temperature, advice is given from an enterprise point of view on how to plan, install and maintain such a measuring campaign.
The second part of the resentation describes sensor networks from an informational point of view. Since the adoption of the Observation and Measurement Standard[1] by the Open Geospatial Consortium (OGC) in 2007 there is a unique conceptual model to describe sensor observations. The author compares the OGC Standard with traditional models and highlights its pros and cons, using a temperature observation as an example.
The last part of the talk is about processing of sensor data. The author introduces simple data mining techniques which may be integrated in service architectures [3] as a sensor control service. His main focus is on the determination of completeness, detection of invalid values (e. g. spikes) and the aggregation of values in different time scales. At the end an example in analyzing a temperature time series using the ZOO Package[7] in R[5] is given.