Sensor Networks

Reference: MOBICOM 2002 Tutorial T5

Wireless Sensor Networks

CSCI 6760
Satya Sanket Sahoo
Outline
 Introduction
 Areas of Application
 The System
 Communications
 Deployment issues
 Energy
 Time & Space
 Collaborations
• The Future
02/14/05 2
Introduction
 Chain Home - Britain’s Radar Network WWII
 Cold War:
 SOSUS – The Pacific Ocean
 NORAD – Cheyenne Mountain
 National Power Grid
 Involves
 Devices with multiple sensors
 Network via wireless/physical links

02/14/05 3
Involved Technologies

Network
Technology

Computational Sensor
Power
Network
Sensor
Technology

02/14/05 4
Application Areas

 Military
 Infrastructure security
 Environment & Habitat Monitoring
 Industrial Sensing
 Traffic Control
 Seismic Studies
 Life Sciences

02/14/05 5
The Systems involved

 Sensor Node Internals

 Operating System
 Physical Size

02/14/05 6
Sensor Node Internals

CPU

POWER
SUPPLY
D
ARE IC ELECTRO-MAGNETIC
T
FR S C
IN OU MI
SENSOR COMMUNICATION INTERFACE
AC EIS GE C…
S A TI
IM NE
AG NODE
M

Some Current Node Platforms:

1. Sensoria WINS
2. Smart Dust – Dust Inc. Berkeley
3. UC Berkeley mote – Crossbow ([Link])

02/14/05 7
Operating System - TinyOS
 Custom built at UC, Berkeley for wireless
sensor nodes
 Component-based architecture: ensures
minimum code size
 Component library includes:
 Network protocols
 Sensor drivers
 Data acquisition tools
 Distributed services

02/14/05 8
Physical Size

WINS Berkley
LWIM III AWAIRS I NG 2.0 Motes

AWACS

02/14/05 9
Communication

 Network Protocol
 Network Discovery
 Network Control & Routing

02/14/05 10
Network Protocol

 For wireless sensor networks: IEEE 802.11

standards
 Personal Area Networks (PAN): IEEE 802.15
standard
 Radius of 5 to 10m
 Ideal application in short-range sensors

02/14/05 11
Network Discovery
 Knowledge of identity and location of its
neighbor
 For ad hoc networks:
 Topology is built in real time
 Periodic updates – sensors fail, new sensors
added
 Global knowledge is generally not needed
 In the absence of GPS, relative positioning
algorithms are used

02/14/05 12
Network Control & Routing
 Network adapts dynamically to
 conserve resources like energy and available
nodes
 Make optimum use of bandwidth and processing
power
 Connectivity must emerge as needed from
algorithms
• Directed Diffusion routing
 Data identity is separate from node identity
 Promotes adaptive, in-network processing

02/14/05 13
Deployment Issues

 Energy Issues
 Time and Space
 Collaboration towards Objective

02/14/05 14
Energy Issues

 Energy drains
 Energy Management

02/14/05 15
Energy Drains
 Processing
 Need for Idle & sleep modes
 Variable voltage & frequency
 Communications
 Currently around 150nJ/bit – short range
 Sensors
 Depends on the sensing modality
 Actuators
 Power Supply

02/14/05 16
Energy Management
 Processor Energy management:
 Shutdown
 Dynamic scaling of frequency and supply voltage
 dynamic scaling of frequency, supply voltage, and
threshold voltage
• Communication Energy management:
 Shutdown based
 Turn off sender and receiver

02/14/05 17
Time and Space

 Sensor node localization

 Temporal synching up
 Scale of operations

02/14/05 18
Localization
 Absolute position on geoid
 e.g. GPS
• Self-positioning vs. Remote-positioning
• Multiple techniques for Location sensing
 Measure proximity to landmarks
 Position relative to initialization point – Dead
reckoning
 Measure distance of landmarks - Ranging

02/14/05 19
Temporal Synchronization
 Reference-broadcast synchronization: Very
high precision sync with slow radios
 Post-facto synchronization: syncs up only
when needed (saves on energy)
 Peer-to-peer synchronization: There are no
master clocks
 Tiered Architecture: Range of node
capabilities

02/14/05 20
Scale – Temporal and Spatial
 Sampling: sampling frequency depends on
 Type of modality being measured
 Type of application
• Extent
• Density: “sensor nodes per footprint of input
stimuli”

02/14/05 21
Collaboration – towards objective
 Requirements – illustrative example
 CSP – Collaborative Signal Processing

02/14/05 22
US Navy - TWMD

02/14/05 23
Collaborative Signal Processing
 Sensor nodes collaborate to extract and
process information – towards common
objective
 Two types of collaboration:
 Intra-node collaboration
 Inter-node collaboration

02/14/05 24
The Research Challenges

 Network deployment and organization

 Query processing and routing
 Storage management
 Increasing the longevity and robustness of
the network
 Security issues
 Address privacy concerns

02/14/05 25
Reference

 Sensor networks: Evolution, opportunities, and challenges Chee-Yee Chong;

S.P. Kumar
 Connecting the Physical World with Pervasive Networks Deborah Estrin, David
Culler, and Kris Pister
 Sensor Networks: Applications and Research Challenges Niki Trigoni Cornell
University
 MOBICOM 2002 Tutorial T5 Wireless Sensor Networks Deborah Estrin & Mani
Srivastava destrin@[Link], mbs@[Link] UCLA Akbar Sayeed
akbar@[Link] University of Wisconsin, Madison

02/14/05 26