Sensor Networks
Download
Report
Transcript Sensor Networks
Taking Sensor Networks from
the Lab to the Jungle
ECE-695
Mobile Wireless Networks
Vamshi Nadipelli
Preethi Tiwari
What is a Sensor Network?
• A sensor network is a collection of sensor nodes
equipped with sensing, communication (short range
radio) and processing capabilities.
02/6/07
2
Outline
Introduction
The System
Areas of Application
Challenges
02/6/07
Technical
System
Conclusion
3
Introduction
Chain Home - Britain’s Radar Network WWII
Cold War:
SOSUS – The Pacific Ocean
NORAD – Cheyenne Mountain
National Power Grid
Involves
02/6/07
Devices with multiple sensors
Network via wireless/physical links
4
Involved Technologies
Network
Technology
Computational
Power
Sensor
Network
Sensor
Technology
02/6/07
5
The Systems involved
02/6/07
Sensor Node Internals
Operating System
6
Sensor Node Internals
CPU
POWER
SUPPLY
SENSOR
COMMUNICATION
ELECTRO-MAGNETIC
INTERFACE
NODE
Some Current Node Platforms:
1. Sensoria WINS
2. Smart Dust – Dust Inc. Berkeley
3. UC Berkeley mote – Crossbow (www.xbow.com)
02/6/07
7
Operating System - TinyOS
Custom built at UC, Berkeley for wireless
sensor nodes
Component-based architecture: ensures
minimum code size
Component library includes:
02/6/07
Network protocols
Sensor drivers
Data acquisition tools
Distributed services
8
Physical Size
LWIM III
AWAIRS I
WINS
Berkley
NG 2.0
Motes
AWACS
02/6/07
9
Applications
Border Monitoring
Battlefield Observation
Forest Fire Detection
Environment and Habitat Monitoring
Infrastructure security
Industrial sensing
Medical Applications
02/6/07
10
Border Monitoring
Most widely cited application
Detection can be based on
sound or vibration
With in a range of 10m
Estimated need of 440,000
sensors
US-Mexico Border (3100 km)
Requires Full length Surveillance
Air dropped biannually (battery life
6months)
Not cost effective
Should distinguish humans from
wild animals
02/6/07
11
02/6/07
12
Battlefield surveillance
Observing enemy activities in a battle field.
Unmanned aerial vehicles (UAV)
Coverage problem (limited radio range)
10,000 nodes were required to monitor just 1 square
kilometer
For large areas:
cost
02/6/07
Many nodes implies large number of UAV’s operating
simultaneously.
13
02/6/07
14
Forest Fire Detection
A sensor network is more feasible as an early warning system for
forests.
Carefully placing nodes (close to vulnerable areas such as hilltops)
Reduce the number of sensors required to cover a large geographic
area.
Important aspect is lifetime
Must operate for a very long period of time to discover a
comparatively rare event
02/6/07
15
Nodes are subjected to random failures
• Due to battery exhaustion
• Disorientation of antennas (falling branches, wind etc)
So, Networks relay messages hop by hop failure of several
closely spaced nodes could partition the network into non
communicating subnetworks
02/6/07
16
Environment and Habitat Monitoring
•Environmental monitoring involves collecting readings over time across a
volume of large space enough to exhibit significant internal variation.
•Environmental sensors are used to monitor relative humidity, barometric
pressure and temperature.
•They study vegetation responses related to climatic trends and diseases
Whereas the imaging sensors can identify, track and measure the population
of birds and other species.
02/6/07
17
Monitoring nesting
Large number of
burrows.
Long time observation
02/6/07
18
Over 100 sensor
nodes.
Long term observation
02/6/07
19
Migration pattern of zebras
They generally move in wide area
Long term observation
02/6/07
20
Sensors were integrated on
to the zebra’s neck.
Consisted of 2 radios.
Long range (base station)
Short range (neighbors)
These were used to monitor
the heart beat, body
temperature and frequency
of feeding
02/6/07
21
Infrastructure Security
• Early detection of chemical, biological and nuclear threats.
• Protection of power plants and communication centers.
• Networks of video, acoustic and other sensors are deployed around these
facilities.
• When compared to Fixed sensors, Ad hoc networks can provide more
flexibility and additional coverage.
• MULTIPLE SENSORS provide Improved coverage, detection, and reduced
false alarm rate.
02/6/07
22
02/6/07
23
Industrial Sensing
Goals of commercial industry
•Lower cost
•Improved performance
•Maintainability
•It involves continuous monitoring of vibrations, lubrication levels and inserting
sensors into regions inaccessible by humans.
•Spectral and Optical sensors are generally used in industrial applications
because inputs from hundreds or thousands of sensors can be fed into the
databases that can be accessed in any number of ways to show the real time
information (called MULTIPOINT OR MATRIX SENSING).
02/6/07
24
Medical applications
Heart rate
Oxygen saturation
Enhances emergency
medical care.
02/6/07
25
Challenges
Power
Communication
Hostile Environments
Cost
02/6/07
26
Technical challenges
Changing network topology:
Node failures
Introduction of additional nodes variations in sensor location
Changes to cluster allocations in response to network demands
requires the adaptability of underlying network structures and
operations.
Advanced communication protocols
To support high level services and real-time operation
(to adapt rapidly to changes in network conditions).
Resource optimization:
To minimize cost, power and network traffic loads
Ensuring network reliability and adequate sensor resolution for data
accuracy.
02/6/07
27
Limitations:
Power, Memory, processing power, life-time. These physical constraints
may be minimized through further technological breakthroughs in
materials and sensor hardware designs.
Failure prone:
Individual sensors are unreliable, particularly in harsh and unpredictable
environments.
Addressing sensor reliability can reduce the level of redundancy
required for a network to operate with the same level of reliability.
Network congestion resulting from dense network
deployment:
The quantity of data gathered may exceed the requirements of the
network and so evaluation of the data and transmission of only relevant
and adequate information needs to be performed.
02/6/07
28
Self-organization
Ability to adapt to dynamic environments as well as ad hoc distribution
and connectivity scenarios.
Self-operating and self-maintaining
This functionality is desired in order to minimize further human
interaction beyond network deployment.
Security
It is a critical factor in sensor networks.
An effective compromise must be obtained, between the low bandwidth
requirements of sensor network applications and security demands.
02/6/07
29
Conclusion
Sensor networks are application specific
Key application characteristics
Lifetime, cost, data rate, environment, network
topology, user interaction
Must address the system aspects of wireless sensor
network design
02/6/07
30
QUESTIONS
?
02/6/07
31