A
Brief
History
of
Sensor
Networks


Muneeb
Ali


Princeton
University

Introduction: 2003

Image:
MIT
TechReview

Introduction: 2005

5000
4400
4500
4000
3400
3500
3000

2500
1960
2000
1500
946
1000
447
500 226
47 43 50 70
0

Image:
Koen
Langendoen

 Introduction: 2008

Censor
Networks:
A
CriOque
of
“Sensor
Networks”
from
a
Systems
PerspecOve


B.
Raman,
K.
Chebrolu
(IIT
Bombay)


Introduction

This
talk:


‐ 
A
brief
history
of
the
last
5
years
of
research
(2003‐2008)


‐ 
Important
problems,
soluOons,
and
lessons


‐ 
Future
direcOons



A Brief History
I
am
smiling

because
I
was

right!


Gordon
Moore

Intel
Co‐Founder

 A Brief History
log
(people
per
computer)


0


1960
 1970
 1980
 1990
 2000
 2010


Image:
Culler:2004

Network Stack

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

L1: Physical

‐ 
Memory:
10
KB

‐ 
Radio:
CC2420
(250
Kbps)

‐
Processor:
MSP430
(16‐bit)


Transceiver

Memory
 Embedded

Processor

Sensors

Ba^ery

L1: Physical
L1: Physical

Intel
Imote
2
 Sun
Spot



●
320/416/520MHz
PXA271
XScale
Processor

 ●
32
bit
ARM7
core








‐
32MB
SDRAM
on‐board

 





‐
256K
RAM








‐
32MB
Flash
on‐board
 





‐
2M
Flash

●
[802.15.4]
Radio
(ChipCon
CC2420)

 ●
[802.15.4]
Radio
(ChipCon
CC2420)


Image
courtesy

 Image
courtesy

L1: Physical

(mm2)
 (MHz)


(kB)


(kB)


(kB)


Reference:
Jan
Beutel,
Metrics
for
Sensor
Network
Plaforms,
ACM
RealWSN
June
2006

Network Stack

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

L4: Applications

Great Duck Island

[Berkeley, 2002]
L4: Applications

Sniper detection

[Vanderbilt, 2003]
L4: Applications

[Princeton, 2004]
L4: Applications

[Del7,
2006]


L4: Applications

[Harvard,
2007]

Network Stack: Challenges

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

Challenges: Energy

<<<
Without

~2
kcal
(per
ba^ery)
 















~280
kcal


cheese!!!












(Quarter
Pounder)


Image:
Koen
Langendoen

Challenges: Wireless

Signal
propagaOon
ranges


sender
 sender


transmission
 transmission


detecOon
 detecOon

distance
 distance

interference
 interference


Image:
Koen
Langendoen

Challenges: Wireless

ReflecOons
/
Shadowing

Challenges: Wireless

[G. Janssen, ET4358]

Challenges: Wireless

2300

2200 RSSI = 1/distα

2100

RSSI
2000
1900

1800

1700
0 5 10 15 20 25 30 35 40
Distance (metre)

Grey
Area
Effect

Challenges: Wireless

Link
layer
&
mulOpath
fading

CC2420 @ 2.4 GHz, power = -1dBm, 2am

0.9
0.8
0.7
100 0.6
% good 50 0.5
0.4 y (lambda)
0 0.3
0 0.1 0.2
0.2 0.3
0.4 0.5 0.1
0.6 0
x (lambda) 0.7 0.8
0.9

[Robert Poor, Ember corp.]

Challenges: RAM

‐
Scalability


‐
Limited
RAM
is
fundamental

‐
Effects
power

‐
O(N)
state
infeasible



Challenges: Applications

The
Internet






v
s.









Sensor‐Nets

Challenges: Summary

The
Internet






v
s.









Sensor‐Nets


●
Independent
hosts
 
●
CollaboraOve
use

●
End
to
end
flows
 
●
Collect,
disseminate,
...

●
Infrastructure
 
●
Ad‐hoc

●
Wired
(generally)
 
●
Wireless


●
Energy

●
Latency,
throughput


●
Bandwidth
is
expensive

●
Bandwidth
is
relaOvely
cheap


Reference:
Philip
Lewis,
ICSI
Talk,
May
2004

Research Problems

‐ 
Medium
Access
Control

‐ 
RouOng


‐ 
LocalizaOon

‐ 
OperaOng
Systems

‐ 
Security

‐ 
Programming
AbstracOons

‐ 
Query
Processing

Network Stack

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

L2: MAC

Contact
person:
Koen
Langendoen


L2: MAC

Classic
Paper:
S‐MAC
(UCLA)

L2: MAC

Read:
MAC
Survey
by
Koen
Langendoen


Network Stack

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

L2/L3: Naming and Addressing

Network
Address
DesOnaOon

|

Next‐Hop
MAC
Address

|
DATA






64
bit
|
64
or
16
bit
|





605
|
109
|





 

605
|
231
|





 

605
|
605
|







921
 109
 231
 605


A

|

4

|





 A

|

9

|





 A

|

2

|







A^ribute
Based
Name

|

Next‐Hop
Locally
Unique

|
DATA







Read:
‘Low‐level
Naming’
paper
(UCLA)


Network Stack

ApplicaOon
Layer

L4


L3
 Network
Layer


L2
 Link
Layer


L1
 Physical
Layer

L3: Traditional View

Sink


Sensornet
Nodes

L3: Traditional View

Internet


Proxy/Sink
 Sensornets

 L3: New View

Ad‐Hoc


Networks


Internet

Personal
Area
Networks

L3: Routing

Sensornet
RouOng
Protocols


Clustering
 MulE‐hop
 Geographic


LEACH
 Directed‐Diffusion
 GPSR

Energy*Delay
 BVR

L3: Routing - LEACH

Image
courtesy
of
Wendi
Heinzelman


L3: Routing - LEACH
L3: Routing - VRR

Node
IDs
are
Integers

(locaOon
independent)

L3: Routing - VRR
Operating Systems

ConOki

MANTIS

SOS


Read:
TinyOS
book
chapter
by
Phil
Levis

Research Problems

‐ 
Medium
Access
Control

‐ 
RouOng


‐ 
LocalizaOon

‐ 
OperaOng
Systems

‐ 
Security

‐ 
Programming
AbstracOons

‐ 
Query
Processing


See
Phil’s
course
at
Stanford
and
Ma^’s
course
at
Harvard

What We Have Learned
Where Do We Go From Here

‐ 
RFID
sensors
(Moore’s
Law)


‐ 
Internet
ciOzens
(1st
class)




‐ 
Urban
Sensing


‐ 
Energy
Management
(Ember)


‐ 
Physical
Sensing
(Industry,
Home)


‐ 
Startups

Further Information

Muneeb Ali
http://muneeb.org

Thank You !