Summary of Activities

Summary of Activities

Workshop on Environment and Sensor Networks

US France Young Engineering Scientists Symposium

Organized with support by

French Embassy

And hosted by

Washington, DC, October 22-24, 2007

Introduction

The goal of this 2.5-day workshop was to bring together young French and US researchers from the disciplines of ecology and information technology (both broadly defined) to establish new contacts that bridge disciplinary and national boundaries, discuss new research challenges and opportunities, seed new interdisciplinary collaborations, and define concrete research activities.

Presentations

The workshop began with 10-minute presentations from the invited researchers followed by

5 minutes for questions and discussion.

Brainstorming Session

Following the presentations, the coordinators facilitated a session in which the participants were invited to contribute to a “gallery of ideas”. The participants were asked to frame their ideas in the context of the following questions, with a focus on key challenges at the interface of sensor network IT and ecological research:

1. 
   What are the essential issues of environment monitoring with sensor networks, and the essential issues for sensor networks in such monitoring activities?
   

2. 
   What is the research which needs to be done, and why?
   

In the following, the generated ideas are gathered into three rough and necessarily overlapping categories: ecological challenges, sensor network design challenges, and IT challenges.

Challenges in the Ecological Sciences

• 
  Animal ecology: density of animals, occurrence of rare animals, trajectories of animals, (feeding behaviours, social behaviours, transmission of diseases, nutrient recycling, elementary sensing).
  

• 
  Climate change: reliable measurement, validation of the measurements, and forecasting. Management, visualization and understanding of spatial and temporal information.
  

• 
  Miniaturization of sensors: high-frequency sampling and on-board processing. Data acquisition from remote sites. Underwater communication.
  

• 
  Image/video sensors for aquatic environments: measurement of nutrients in both terrestrial and aquatic systems.
  

• 
  Statistical modelling/learning of data to understand the patterns.
  

• 
  Heterogeneous sensor networks; scaling in time and space.
  

• 
  Nanosensors (in the sea, in the air, in the soil).
  

• 
  Databases of existing sensor data.
  

• 
  Rules/protocols for setting up sensor networks.
  

• 
  Smart/adaptive sensors that adjust to or filter the data.
  

• 
  Standardized data format and protocols.
  

• 
  Enhancement of existing data loggers with networking components.
  

Challenges in Sensor Network Design

• 
  Data processing (low power on-board processing, data cleaning, compression), and out-of-network
  
• 
  processing, (challenge: huge amount of data). Automate this process.
  

• 
  Trade: aggressive data processing and compression vs. data validation, (the raw data is lost and how much do we trust the correctness of the compressed data)?
  

• 
  Data reliability, (loss of sensors). Employ coding to enhance the data robustness. Challenges: data coherence, ensure data recovery, power consumption for data exchange of sensors.
  

• 
  Data retrieval for static large scale sensor nets; data mules (vehicles, airplanes, humans with portable devices such as mobile phones). Challenges: communication bandwidth and power constraints on sensor nodes?
  

• 
  Energy efficiency in untethered sensors.
  

Challenges in Information Technology

• 
  Develop abstractions that are similar across different ecological and environmental applications and disciplines; classify/categorize the applications.
  

• 
  Systematic data sharing and data standardization.
  

• 
  Data processing, replication, retrieval. Compressed sensing and adaptive sampling (to conserve energy, reduce sampling rate, simplify data collection, and reduce bandwidth requirements). Classification problems.
  

• 
  Localization in environments where GPS is not available.
  

• 
  WiMAX-based communication techniques (medium/long-haul communication).
  

• 
  Well-engineered systems composed of usable wireless sensing nodes. Current commercial products do not yet work easily.
  

• 
  Need for representative data sets, network activity traces, and innovative test environments.
  

• 
  Hardware platforms, (maybe tuneable to applications).
  

• 
  Data architecture/management; data integrity and its management.
  

• 
  Scalability and reliability.
  

Application-Focused Small Group Meetings

After the brainstorming session, the consensus of the group was that the ecological and environmental sensing applications were very broad, with diverse sensing/IT requirements. In addition, the coordinators addressed difficulties of the participants in indentifying research challenges linking environmental and IT research.
To identify and elaborate specific key research challenges, the ecological researches participants self-organized into three areas exemplified by three case studies. In the following, the needs and challenges identified by the three groups are summarized.
Understanding of Animal Movement and Social Behaviours

• 
  Collar systems for, e.g., badgers
  
• 
  3D accelerometers for dead reckoning of animal movement
  
• 
  Adaptive/intelligent sampling
  
• 
  Passive systems (e.g., RFID) for data collection
  
• 
  Measurement of interaction: Identify the patterns/activities of the animals of interest to the ecologist from the motion trajectories of the animals; identify the contact patterns.
  
• 
  Measurement of physiological processes, e.g., heart rate
  
• 
  Measurement of environmental conditions
  
• 
  Dissemination of data between sensor nodes to help deliver the data to the base station.
  

Spatial Variation in Terrestrial Net Primary Productivity

• 
  High density spatial sampling and determination of spatial/temporal variability
  
• 
  Example system: flux tower site
  
• 
  Measurement of light, soil moisture, relative humidity, temperature
  
• 
  Sampling frequency/interval scales from 20 Hz to 30 minutes depending on variable
  
• 
  Adaptive sampling of spatial and temporal components of the network
  

Aquatic Systems - Physical Structure of the Environment and Net Primary Productivity

• 
  Continuous monitoring of physical and biological variables over depth
  
• 
  ~1 sample/hour baseline sampling interval; remote modification of sampling schedule
  
• 
  real-time data at desktop (or at least daily updates)
  
• 
  image/video sensing (microorganisms)
  
• 
  seasonal temporal scales
  
• 
  Under-water communication schemes that provide sufficient bandwidth and energy efficiency
  

Recommendations for Further Activities

After the three ecologically-focused groups reported on needs and challenges, the IT researchers were invited to join the group whose challenges aligned most closely with their research capabilities and goals. The three groups then were charged with the goal of defining research ideas and associated projects that they will continue to develop following the workshop, with the ultimate goal of writing international/interdisciplinary grant proposals.

Animal Behaviour

Identify: Problem or Issue	Assess the physiological, behavioural and ecological processes by which animal organisms and populations interact with their environment
Analyze: Root Cause	Get accurate, unbiased, reliable data from miniaturized low-powered high-frequency samplers simultaneously deployed on multiple free-moving animals
Main research direction	Develop an opportunistic wireless network and data management system of low-powered high-frequency sampling heterogeneous mobile sensors
Research ideas / project ideas		Participants
Identify network architecture in disconnected operation Keep contact with animals that cannot be easily re-captured		Frédéric WEIS, INRIA Rennes, France Marc Olivier KILLIJIAN, LAAS Toulouse, France
Compressive sensing of high-rate signals, post-facto data compression Conserve energy and memory when recording huge data (ECG, 3D acc)		Nirupama BULUSU, Portland State, USA Jie GAO, Stony Brook University; USA
Low power remote data collection Up/download data from loggers to PC and base station		Michaël HAUSPIE, INRIA Lille, France
Analysis and feature extraction from mobility data Treat complex data to extract biological patterns		Jie GAO, Stony Brook University; USA
Network organisation Assess individual localisation and social interactions		Guillaume CHELIUS, INRIA Lyon, France
Tracking-based data fusion of GPS and accelerometer data Mix and adjust location data issued from different methods		Nirupama BULUSU, Portland State, USA Michaël HAUSPIE, INRIA Lille, France
Infer biological and environmental processes from bio-logging data		Tanguy DAUFRESNE, INRA Toulouse, France Jean-Yves GEORGES, CNRS Strasbourg, France

High-Density Sensing of Environmental Variables

Identify: Problem or Issue	How to capture spatial variability in biologically relevant parameters, e.g., in eddy flux tower footprints
Analyze: Root Cause	Parameters that drive primary productivity are highly variable spatially and temporally and Eddy Flux techniques integrate measurements over large areas
Main research direction	High density environmental monitoring using adaptive network to interpret primary productivity data
Research ideas / project ideas		Participants
Adaptive sampling (e.g. in response to changing wind direction/speeds)		Amol Deshpande, Louis Santiago, Marcy Litvak, Yonghe Liu, Robert Hollister, J.-C. Clement, Jerome Chave
Customized sensor nodes (1. Harsh conditions, 2. Many sensors per mote, 3. Sat-flow sensors)
Hydraulic regulation of primary productivity
Efficient networking for long term monitoring, data quality issues (noise, cleaning up, etc.)

Aquatic Monitoring

Identify: Problem or Issue	Remotely measure suite of physical (T, S, PAR) and bio/chemical (nutrients, O2, chlorophyll fluorescence, pH, alkalinity) parameters in diverse aquatic systems.
Analyze: Root Cause	Cannot resolve vertical and temporal variability in the water column with single hydrocasts/available sensors. Cannot cover large spatial areas. Long time series.
Main research direction
Research ideas / project ideas		Participants
Detecting changes in plankton dynamics along large environmental gradients (Latitudinal, nutrient) in lacustrine environments as a response to global forcing.		Robinson, Litchman, de Garidel, Levis
Assessing the impact of coastal nutrient loading on estuarine and open ocean systems.		Robinson, Litchman, de Garidel, Levis
Optimization of seasonal sampling of lacustrine and marine calcareous plankton and assessment of its response to long term acidification		Robinson, Litchman, de Garidel, Levis

Unifying Themes

• 
  There are clear needs and research challenges that link the Animal Behaviour and Spatial Variability of Net Primary Productivity thrust areas.
  
• 
  There is currently a lack of mechanisms for information exchange between the disciplines.
  
  • 
    Information sources on IT capabilities in environmental sensing should be developed.
    
• 
  There is a strong need for environmental/ecological test beds and real-world datasets.
  
• 
  Data management/standards should be defined across applications (US NEON project can be a resource here).
  
• 
  Funding agencies should promote research across experiments and scales.
  

Substantial progress was made during this workshop, as there were no collaborations between participants from sensors and environment before this workshop. However to make this workshop totally successful, concrete research proposals should be developed targeting research programs at US, French, and EU agencies (e.g, NSF, European Commission, ANR,…).

Participants

Nirupama Bulusu, Portland State University, Department of Computer Science, Maseeh College of Engineering and Computer Science, P.O. Box 751, Portland, OR 97207-0751.
Jérôme Chave, CNRS-UPS, Laboratoire Evolution et Diversité Biologique (UMR 5174), 118 route de Narbonne, 31062 Toulouse Cedex.
Guillaume Chelius, INSA de Lyon, Laboratoire CITI, Bâtiment Léonard de Vinci, 21 avenue Jean Capelle, 69621 Villeurbanne.
Jean-Christophe Clément, Université de Grenoble, UMR 5553 Laboratoire d'Ecologie Alpine (LECA), BP 53, 2233 rue de la Piscine, 38041 Grenoble Cedex 9.
Tanguy Daufresne, INRA-CEFS, Chemin de Borde Rouge, BP 52627, 31326 Castanet-Tolosan Cedex.
Thibault de Garidel-Thoron, CEREGE-CNRS, Europôle de l'Arbois, BP 80, 13545 Aix-en-Provence, Cedex 4.
Amol Deshpande, 3221 A. V. Williams Bldg., Department of Computer Science, University of Maryland, College Park, MD 20742.
Jie Gao, Stony Brook University, 1415 Computer Science Bldg., Stony Brook, NY 11794.
Jean-Yves Georges, CNRS, Institut Pluridisciplinaire Hubert Curien (UMR 7178), Departement Ecologie, Physiologie et Ethologie, 23 rue Becquere, 67087 Strasbourg Cedex 2.
Michaël Hauspie, IRCICA, Parc Scientifique de la Haute Borne, 50 avenue Halley,BP 70478, 59650 Villeneuve d'Ascq.
Robert Hollister, Grand Valley State University, Department of Biology, 209 Henry Hall, 1 Campus Drive, Allendale, MI 49401-9403.
Marc-Olivier Killijian, LAAS-CNRS, 7 avenue du Colonel Roche, 31077 Toulouse Cedex 4.
Philip Levis, Stanford University, 358 Gates Hall, Stanford, CA 94305-9030.
Elena Litchman, Michigan State University, Department of Zoology and Ecology, Evolutionary Biology and Behavior Program, W. K. Kellog Biological Station, 3700 East Gull Lake Drive, Hickory Corners, MI 49060.
Marcy Litvak, University of New Mexico, Department of Biology, 167 Castetter Hall, MSC03 20201, Albuquerque, NM 87131-0001.
Yonghe Liu, The University of Texas at Arlington, Department of Computer Science and Engineering, Box 19015, 416 Yates Street, Rm 305 Nedderman Hall, Arlington, TX 76019-0015.
Guillaume Marrelec,
Rebecca Robinson, University of Rhode Island, Graduate School of Oceanography,Narragansett Bay Campus, Narragansett, RI 02882.
Louis Santiago, University of California, Department of Botany and Plant Sciences, Batchelor Hall 4113,Riverside, CA 92521.
Frédéric Weis, IRISA, Campus Universitaire de Beaulieu, 35042 Rennes Cedex.

Scientific Coordinators

■ Michel Banâtre, Directeur de Recherche INRIA, Project team: Ambient Computing and Embedded Systems (ACES), Rennes.

■ Kevin Griffin, Lamont-Doherty Earth Observatory of Columbia University, New-York

■ Paul Flikkema, College of Engineering and Natural Sciences, Northern Arizona University, Flagstaff