Wireless Sensor Network Research and

Experimental Experience

David Q. Liu, Ph.D.

Dept. of Computer Science and Engineering
The Ohio State University
February 18, 2004
 Outline
• Introduction
• Wireless Sensor Network Architecture
• Network Layer Protocols
• LITeS and Echelon Projects at OSU
• Conclusion
 Sensor Networks
• A sensor network is composed of a large number
of sensor nodes densely deployed either inside the
phenomenon or very close to it: sensing,
communication, and data processing
• Features:
– Random deployment
– Self-organizing
– Cooperating
– Local computation
 Applications
• Military
– Command, control, communication, intelligence,
surveillance, reconnaissance, targeting systems
• Health
– Patient monitoring, disable patient helping
• Commercial
– Inventory management, product quality monitoring,
disaster area monitoring
 Characteristics
• Large number of nodes (versus ad hoc networks)
• Densely deployed
• Prone to failure
• Frequent topological changes
• Broadcast communication (versus point-to-point)
• Limited power, computation, and memory
• No global identification (ID)
Sensor Networks
Sensor Node
Berkeley TinyOS Motes
 Design Factors
• Fault Tolerance
– Power failure, physical damage, environmental
interference
– Reliability Rk e k t
• Scalability
– Hundreds, thousands, even millions
– Density  ( R) ( N R 2 ) / A
• Product Costs
– Cheaper than traditional sensors
 Design Factors
• Hardware Constraints
– Smaller size
– Extreme low power consumption
– High volume density
– Dispensable and autonomous
– Unattended
– adaptive
• Sensor Network Topology
– Deployment (hand-placed, plan dropping,
rocket/missile delivery
– Position changes, reliability, available energy,
malfunctioning, task dynamics
– Additional nodes
 Design Factors
• Environment
– Unattended remote geographic areas
– Inside a large machinery at the bottom of an ocean
– Contaminated field
– Battlefield
– Home or large building
• Transmission Media
– Radio, infrared, optical media
 Design Factors
• Power Consumption
– A limited power source (<0.5 Ah, 1.2 V)
– Battery life
– Power conservation and power management
Sensor Network Protocol Stack
 Network Layer Protocol
• Design Principles
• Network Schemes and Routing Protocols
 Design Principles
• Power Efficiency
• Data-Centric
• Data Aggregation
• Attribute-based addressing and location
awareness
 Power Efficiency of Routes
Choices:
- Max PA
- Min 
Route 4: T-F-E-Sink, total PA = 5. Total  = 6
- Min hops
- MaxMin PA

Route 1: T-B-A-Sink, total PA = 4, total  = 3

Route 3: T-C-D-Sink, total PA = 3. Total  = 4

Route 2: T-C-B-A-Sink, total PA = 6. Total  = 6

 Data Centric Routing
• Assign the sensing tasks to the sensor nodes
– Sink broadcast the interest
– Sensor nodes broadcast an advertisement
• Attribute-based naming
– “The areas where the temperature is over 70º F”
versus
“the temperature read by a certain node”
Data Aggregation
 Network Schemes and
Routing Protocols
• Small Minimum Energy Communication Network
(SMECN)
• Flooding
• Gossiping
• Sensor Protocol For Information via Negotiation
(SPIN)
• Sequential Assignment Routing
• Low-Energy Adaptive Clustering Hierarchy
(LEACH)
• Directed Diffusion
SPIN Protocol
Direct Diffusion
 Network Layer Protocol Issues
• Higher topology Changes
• Higher Scalability
• Location Awareness
OSU Research and Experiments
• DARP (Defense Advanced Research
Program)’s NEST (Networked Embedded
Software Technology ) Program
– Self-Stabilization in NEST (2001)
– Line in the Sand (LITeS) (2003)
• how to detect, classify, and track various types of objects (such
as persons and cars) using many, resource-poor smart dust
sensor nodes (~100).
– Echelon (Extreme Scaling) (2003 - 2004)
• to investigate the challenges in scaling to a network of 10,000
sensor nodes.
 Military Objective
• Given a relatively large, ad hoc
perimeter/border,
– use a mote network in open, denied areas
– to extend/support detection, classification
& tracking of intruders
– with good cost performance
Concept of Operations
 LITeS Military Problem Addressed
• Improve extant carefully-placed unattended ground sensors
by:
– dealing with extended/open/denied areas
– reducing need for careful placement of sensors & radio repeaters
– automatic instrumentation & remote monitoring and control
• Metrics:
– Cost (dollars per unit area protected)
– Power (watts per unit area protected)
– Robustness (no single point of failure)
– Operational
• ease of distribution to denied areas
• ease of concealment
• ease of deployment effect on force reduction
 Research Objective
• Demonstrate that NEST-middleware
enables a solution to the military objective
in a manner that is
– robust (i.e., tolerates uncertain
environment)
– accurate (i.e., low false negatives & low false
positives)
– cost effective (i.e., using middleware on dense
set of cheap sensors
vs. sparse set of resource rich sensors,
e.g. Steel Eagle or REMBASS)
 LITeS Experiments
• A dense mote, resource poor solution for collaborative
detection, classification & tracking
– coherent & incoherent coordination
– multi-modal sensing
– sample concept: each intruder type has unique influence field
• By way of example, the demo involves
– 500’ long line/perimeter, with 1 or 2 relays to outside network
– detect different types of intruders (vehicles, and persons)
– locate with modest accuracy (1-5 meters)
– track at least one intruder of each type, maintain approximate
count
 Fault Tolerance
• several motes in a region are turned off, and then on again
• some mote locations are swapped
• some motes are displaced from their location
 Performance Requirement
• Probability of detection > 95%
• Probability of false alarm < 1 %
• Detection latency < 10 s
• A vehicle misclassified as a person = zero
• A soldier misclassified as a person < 1%
– Goal: Minimize false positives and false
negatives
 Testing Diagram

Border Landmark Sensor WSN Gateway

Repeater Radio
 Intruders
• Intruders are of one of the following types:
– soldier (person carrying metallic
objects)
– Tank
– Person
• Intruders may be assumed to generally maintain
constant heading and speed-range
– aggregate information is assumed to suffice
 Sensors

Enclosed mote with a magnetometer sensor Enclosed mote with a MIR sensor
 Middleware Services
Visualization Scheduler

• Ad hoc network formation, routing

• Time synchronization Matched Filter Snapshot

• Local matched filter Routing

• Regional matched filter

Power Mgmt

• Snapshot Localization

• Visualization
• Sensor calibration Time Sync

Sensor Cfg

• Localization
Sensors
• Data aggregation
• Power management 1 1

* *

Magnetometer Radar
 Freeze/Unfreeze
the Visualization

Start/Stop the
Toggle to Turn Playback the
Simulation
Topology Display Simulation (Enabled
On/Off once the simulation is
“frozen”)
Mote Statistics
 Lines showing Parent-
Child Relationships
Magnified Mote (Orange end points to
with Readings parent)
Display

Magnified Target
with Type/Speed
etc.
 Echelon Project
What is ech·e·lon ('e-sh&-"län)?
noun:
1. A DARPA-funded research project in the NEST
Program that seeks to model, design, build, field, and
test the world's largest sensor network consisting of
over 10,000 nodes.
2. A formation.
a. A formation of troops in which each unit is positioned
successively to the left or right of the rear unit to form an
oblique or step like line.
b. A flight formation or arrangement of craft in this manner.
c. A similar formation of groups, units, or individuals.
 Echelon Project
3. A subdivision of a military or naval force.
4. A level of responsibility or authority in a
hierarchy; a rank.
5. A diffraction grating consisting of a pile of
plates of equal thickness arranged stepwise with
a constant offset.
6. (Mil.) An arrangement of a body of troops when
its divisions are drawn up in parallel lines each
to the right or the left of the one in advance of it,
like the steps of a ladder in position for
climbing.
 Echelon Project
verb:
1. To arrange or take place in an echelon.
What about Echelon Project
- group sensors in clusters
- provide the basis for a tactical military tool
- have hierarchical communications and control
- place nodes regular grid or in parallel lines
- large number of nodes (> 10,000)
 Conclusions
• Introduction
• Wireless Sensor Network Architecture
• Network Layer Protocols
• LITeS and Echelon Projects at OSU
• Conclusion
Questions?

Contact Information
David Q. Liu
The Ohio State University
[email protected]