SensorDB: a virtual laboratory for the integration, visualization and analysis of varied biological sensor data.

Dimitrios Georgakopoulos,Scott C Chapman,Doug Palmer,Pablo Rozas-Larraondo,Edward Holland,Robert T Furbank,David M Deery,Jose Jimenez-Berni,Ali Salehi

doi:10.1186/s13007-015-0097-z

Abstract

BackgroundTo our knowledge, there is no software or database solution that supports large volumes of biological time series sensor data efficiently and enables data visualization and analysis in real time. Existing solutions for managing data typically use unstructured file systems or relational databases. These systems are not designed to provide instantaneous response to user queries. Furthermore, they do not support rapid data analysis and visualization to enable interactive experiments. In large scale experiments, this behaviour slows research discovery, discourages the widespread sharing and reuse of data that could otherwise inform critical decisions in a timely manner and encourage effective collaboration between groups.ResultsIn this paper we present SensorDB, a web based virtual laboratory that can manage large volumes of biological time series sensor data while supporting rapid data queries and real-time user interaction. SensorDB is sensor agnostic and uses web-based, state-of-the-art cloud and storage technologies to efficiently gather, analyse and visualize data.Conclusions Collaboration and data sharing between different agencies and groups is thereby facilitated. SensorDB is available online at http://sensordb.csiro.au.

Highlights

To our knowledge, there is no software or database solution that supports large volumes of biological time series sensor data efficiently and enables data visualization and analysis in real time
Field based agricultural, forestry and ecology research studies are often undertaken in remote locations and require the collation of varied data types including: time series data from wireless sensor networks; spatial data from imaging devices; human observations scored and recorded on paper or on a portable tablet device; destructive samples and harvests taken from the field and analysed in a laboratory
Data upload In order to upload sensor data or metadata values to SensorDB, we provide three upload mechanisms: 1. Global Sensor Networks (GSN) virtual sensor GSN [3] is a sensor data processing engine, designed to capture and process realtime data streams

Summary

Introduction

There is no software or database solution that supports large volumes of biological time series sensor data efficiently and enables data visualization and analysis in real time. Forestry and ecology research studies are often undertaken in remote locations and require the collation of varied data types including: time series data from wireless sensor networks; spatial data from imaging devices; human observations scored and recorded on paper or on a portable tablet device; destructive samples and harvests taken from the field and analysed in a laboratory. Such data is typically collated in unstructured repositories on an individual researcher’s computer or on a centrally managed networked file system. Such unstructured data repositories typically do not support data analysis and visualisation for rapid, initial checks of data integrity and are ill suited to large time series data

Objectives

Results

Conclusion