Sensor Networks and Environmental Monitoring

1
Sensor Networks and Environmental Monitoring
Jacques Panchard
2
What is a Sensor Network?

• Ad hoc network (self-organized)
• Environment-to-person communication
• Nodes
• large number (100-1000s)
• sensing capability
• communication, processing, memory capabilities
• energy constrained
• mobile/immobile
• Base station - BS
• energy unconstrained (if connected to the power
  grid)
• mobile/immobile

3
The network is the sensor

• Requires robust distributed systems of
• thousands of
• physically-embedded,
• unattended,
• and often untethered,
• devices.
• (Manges Smith, Oakridge Natl Labs, 10/98)

4
What is a Sensor Network?

• sensor
• aggregator
• relay

5
Why a Sensor Network?

• Possible tasks
• data gathering
• seismic
• acoustic
• medical
• intelligence
• monitoring
• climate
• equipment
• movement

• Possible users
• Authorities
• disaster prevention
• pollution monitoring
• law enforcement...)
• Scientific community
• create or validate models
• Economy
• Agriculture
• Industry
• Services
• Private (home automation)
• Quality of life improvement
• Energy saving

6
Sensor Networks design issues

• Power efficiency
• Node life-time
• Network life-time
• Especially important for self-powered nodes
• Localization
• Range based (Time Of Flight, RSSI etc.)
• Range free (based on anchors with known positions
  and hop count)
• GPS is often impractical because it is energy
  hungry
• Security
• Secure Positioning
• Data Validation
• Packaging
• Strong environmental constraints depending on the
  application
• Data Processing

7
Alternatives to sensor networks

• Remote sensing (satellite)
• Cellular networks (GSM, GPRS...)
• Individual sensing and manual data gathering

8
The case for environmental monitoring

• "The UK Government's chief scientist now says
  climate change is a far worse danger than
  international terrorism" (BBC News, 10 Jan 04)
• Environmental phenomena are a major scientific
  and societal concern
• Pollution
• Water
• Climate
• Information technology, communications and
  sensing are enabling technologies
• Due to current technological changes, there will
  be a watershed phenomenon in
• Ability to monitor
• Price point
• Infrastructure
• Data processing

Courtesy of Prof. Martin Vetterli
9
Monitoring situation
Today 100k / point
5-10 years horizon 10-100 /

• Several orders of magnitude difference in size
  and price!

Courtesy of Prof. Martin Vetterli
10
Implications for science and society

• On the science side
• Completely new datasets are obtainable
• (sensor networks as new telescopes)
• New scientific questions can be addressed
• Scientific theories can be verified/validated
• Measurement and simulation can be tuned
• On the societal side
• Regulators have access to more data
• Closing the loop sensing and actuating
• Communities can become involved

Courtesy of Prof. Martin Vetterli
11
Case studies

• Air pollution monitoring (static and dynamic
  scenarios)
• Water/climate monitoring
• Ground dynamics (glaciers, moving grounds,
  avalanches etc.)

Courtesy of Prof. Martin Vetterli
12
A platform for environmental monitoring the
Mica2 motes

• Made by Crossbow Technology Inc
• Designed for sensor networks
• Atmel ATMega128 processor
• Chipcon CC1000 radio

• Runs TinyOS from its Flash Memory
• 51-pin connector to interface itself
• Add-on sensor boards provide sensing for
  Temperature, Humidity, etc.

5
13
The Operating System tinyOS

• TinyOS created at UC Berkeley, now at
  Sourceforge
• Library includes network protocols distributed
  services
• Also includes sensor drivers and DAQ tools
• Powerful OS, Large community support

6
14
TinyOS Programming Model

• Event-driven, concurrent OS
• nesC supports the TinyOS concurrency model
• Applications Built of components connected by
  well defined
• bi-directional
  interfaces
• Components Form the executable by linking
  together.
• Use Provide
  Interfaces
• Interfaces Declare commands events

Sample Dependency Graph of Components
The Concurrency Model

• Executes only one program
• Two Threads Tasks Hardware Event Handlers
• Tasks Functions that run to completion
• H.E.H Response to hardware interrupt
• Only H.E.Hs may pre-empt tasks or other H.E.Hs
• Data race conditions are highly probable

7
15
COMMON-Sense Net
Agriculture and water management in rural India
with the use of wireless sensor networks
Jacques Panchard (IP10) with Profs. Jean-Pierre
Hubaux and Yves Pigneur
16
Water scarcity today
Water supply, distribution of unserved populations
Sanitation, distribution of unserved populations
Source UNESCO World Water Assessment
Programme, 2003

• Consequence Growing humanitarian crises and
  political instability

17
Water and agriculture

• Developing countries amount of water withdrawn
  for irrigation is more than twice the actual
  irrigation requirements

Source FAO Land and Water development
division, 2003
Agriculture is largely responsible for ground
waters depletion and salinization.
18
Research Challenges

• Test the relevance of the ad-hoc sensor networks
  model in a concrete application
• Assess the use of environment-to-person ICTs in
  developing countries
• Challenge the ad-hoc networking com-munitys
  assumptions on environmental monitoring (focus on
  agriculture)
• Raise and address new technical challenges based
  on a concrete implementation

Field
Social Sciences
EnvironmentalSciences
ICTS
19
A concrete test case
20
A concrete test case (2)

• Pavagada taluk 50 villages over a radius of
  25km. Around 100000 inhabitants
• Chennakeshava Pura around 2000 inhabitants
• Semi-arid climate
• Marginal farmers (lt 1 ha) and small farmers (lt 2
  ha)
• No powered irrigation
• Culturesgroundnut (for oil), cereals trees
  areca nut, coconut...rice in some irrigated
  patches

21
Chennakeshava User survey

• We conducted a survey in three villages of
  Karnataka in April-August 2004.
• The population was segmented in several user
  groups farmers (irrigated, non-irrigated),
  traders, shepherds, labourers.
• It was conducted by local people, asking open
  questions about  What are your information
  needs?
• Key findings Rao 2004
• Information needs are of two kinds
  person-to-person and environment-to-person.
• They differ but overlap between the different
  segments of the population

Rao 2004 PR Seshagiri Rao, Information needs
for farming and livestock management in semiarid
tracts of southern India, Technical report, 2004
22
Survey results
Irrigated Farmers
Non irrigated Farmers
23
Data requirements
24
An heterogeneous sensor network
25
System characteristics

• Self-organizing network of heterogeneous wireless
  sensor-nodes (ease of deployment,
  non-intrusiveness)
• Self-powered and autonomous
• Nodes communicate in a multihop fashion
• Low data-rate
• Dynamic scalability
• Node failure detection
• Internet-connectivity

26
Consortium and resources
27
Consortium a historical perspective
EPFL/SDC Cooperation Fund
Chennakeshavapura Trust (NGO) S. Rao
SDC
IISc
EPFL
HYDRAM Prof. Mermoud
CAOS Prof. Gadgil
MICS IP10
LCA Prof. Hubaux, J. Panchard
CEDT Prof. Jamadagni
UNIL
HEC Prof. Pigneur
28
Project deliverables

• Environmental monitoring sensor network
• User requirements for environmental monitoring in
  rural developing area
• Study on relevance of sensor networks for
  environmental monitoring (focused on developing
  countries)
• Impact assessment
• Business plan and deployment scenario (replicable
  and scalable)
• Publication of papers on technical challenges
• For updated information http//commonsense.epfl.c
  h