Delay Analysis of Large-scale Wireless Sensor Networks

foamyflumpΚινητά – Ασύρματες Τεχνολογίες

21 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

64 εμφανίσεις

Delay Analysis of Large
-
scale Wireless
Sensor Networks

Jun
Yin, Dominican University, River Forest, IL, USA,

Yun Wang, Southern Illinois University Edwardsville, USA

Xiaodong Wang,
Qualcomm Inc. San Diego, CA, USA


1

Outline


Introduction


Delay analysis


Hop count analysis


One

dimensional


Two

dimensional


Source


destination delay analysis


Random source

destination


Delay from multi
-
source to sink


Flat architecture


Two
-
tier architecture


Conclusion

1
-
3

“Cool” internet appliances

World’s smallest web server

http://www
-
ccs.cs.umass.edu/~shri/iPic.html

IP picture frame

http://www.ceiva.com
/

Web
-
enabled toaster +

weather
forecaster

http://news.bbc.co.uk/2/low
/science/nature/1264205.st
m

Internet phones

Wireless Sensor network :
The
next big thing after Internet


Recent technical advances have enabled the
large
-
scale deployment and applications of
wireless sensor nodes.


These
small in size, low cost, low power sensor
nodes is capable of forming a network without
underlying infrastructure support.


WSN is emerging as a key tool for various
applications including home automation, traffic
control, search and rescue, and disaster relief
.

Wireless Sensor Network
(WSN)


WSN is a
network

consisting of hundreds or
thousands of
wireless

sensor

nodes, which
are spread over a geographic area.


WSN has been an emerging research topic


VLSI


Small in size, processing capability


Wireless


Communication capability


Networking


Self
-
configurable, and coordination

WSN organization


Flat vs. hierarchical


Homogenous vs. Heterogeneous

Delay is important for WSN


It determines how soon event can be
reported.


Delay is determined by numerous
network parameters: node density,
transmission range; the sleeping
schedule of individual nodes; the
routing scheme, etc.


If we can characterize how the
parameters determine the delay, we
can choose parameters to meet the
delay requirement.


7

Outline


Introduction


Delay analysis


Hop count analysis


One

dimensional


Two

dimensional


Source


destination delay analysis


Random source

destination


Delay from multi
-
source to sink


Flat architecture


Two
-
tier architecture


Conclusion

Our approach


Firstly, we try to characterize how network
parameters such as node density,
transmission range determine the hop
count;


Then we consider typical traffic patterns in
WSN, and then characterize the delay.


Random
source to random destination


Data aggregation in
two
-
tier clustering
architecture

Outline


Introduction


Delay analysis


Hop count analysis


One

dimensional


Two

dimensional


Source


destination delay analysis


Random source

destination


Delay from multi
-
source to sink


Flat architecture


Two
-
tier architecture


Conclusion

Modeling


Randomly
deployed WSN is modeled
as:


Random geometric graph


2
-
dimensional Poisson
distribution


Nodes are deployed randomly.


The probability of having
k
nodes located with in
the area of around the event :




2
s
r

Shortest path routing: One
dimensional case


At each hop, the next hop is the farthest
node it can reach.

0
r
L














0
]
[
1
]
[
r
e
r
P
r
P










0
]
[
r
e
r
P


0
1
]
[
0
r
e
r
r
E





:
Transmission range

r:
per
-
hop progress








)
(
r
E
L
H
0
r
12

Two
-
dimensional case


Per
-
hop progress

0
r
1
r
1


2

2
r
14
/50

Average per
-
hop progress in 2
-
D case



2
2
0
]
[
1
]
[













r
e
P
P


2
2
0
2
]
[









r
e
P











0
0
0
cos
]
[
]
[
r
d
d
r
P
r
E
Average per
-
hop progress as node
density increases

Numeric and simulation results

Hop count between fixed S/D distance under
various transmission range

It shows that our analysis
can provide a better
approximation on hop
count than .

0
r

15

Hop count simulations

Hop count between various S/D distance

It shows that our
analysis can provide a
better approximation on
hop count than .

r

Outline


Introduction


Delay analysis


Hop count analysis


One

dimensional


Two

dimensional


Source


destination delay analysis


Random source

destination


Delay from multi
-
source to sink


Flat architecture


Two
-
tier architecture


Conclusion

Per
-
hop delay and
H

hop delay


In un
-
coordinated WSN, per
-
hop
delay is a random variable between 0
and the sleeping interval (
T
s
)
.


Per
-
hop delay is denoted by
d
:

2
)
(
s
T
d
E





s
T
s
s
T
ds
T
d
E
s
d
0
2
2
12
1
)]
(
[
)
(

Random source/dest traffic

Hop count between random S/D pairs












2
2
2
4
)
(
2
2
4
/





L
L
L
P
D
S
Distance distribution
between random S/D
pairs in a square area of
L*L:

19

Heterogeneous WSN


Sensor nodes might have different
capabilities in sensing and wireless
transmission.

http://
intel
-
research.net/berkeley/features/tiny_db.asp

Random deployment of
heterogeneous WSN

N
1
= 100

N
2
= 300

L = 1000m

21

22
/50

Modeling


The deploying area of WSN: a square of
(L*L).


The probability that there are
m

nodes
located within a circular area of is:




Node density of Type I and Type II nodes:




,
*
1
1
L
L
N


L
L
N
*
2
2


2
!
)
(
)
,
,
(
2
r
m
e
m
r
r
m
P





2
r

2
-
tier structure

Clusterhead
Type II node chooses the closest Type
I node as its clusterhead:

Voronoi diagram

23

24
/50

Distance distribution

PDF of the distance to from Type II
sensor node to its clusterhead

2
1
1
2
)
(








e
v
P
Distance distribution
between a Type II sensor
node to its closest Type I
sensor node:

1
2
)
(


v
E
Average distance:

Average delay in 2
-
tier WSN





1
2
0
2
0
2
0
)
,
,
(
)
,
,
(
)
(
2
|
)
(





r
F
T
dv
r
F
v
v
P
T
h
H
d
E
E
D
E
s
L
s





Average delay:

Per
-
hop progress

25

26
/50

Summary on delay analysis


The relationship between node density,
transmission range and hop count is
obtained.


Per
-
hop delay is modeled as a random
variable.


Delay properties are obtained for both flat
and clustering architecture.

27
/50

Conclusion


Analysis
delay property in WSN;


It covers typical traffic patterns in
WSN;


The work can provide insights on
WSN design.


Thanks.


Questions?



28

Random source to central sink
node

Laptop computer
29

Incremental aggregation tree

30

31

Hop count analysis (Key
assumptions)