Delay Analysis of Large-scale Wireless Sensor Networks Jun Yin, Dominican University, River Forest, IL, USA, Yun Wang, Southern Illinois University Edwardsville,

1

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

Outline

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

IP picture framehttp://www.ceiva.com/

Web-enabled toaster +weather forecasterhttp://news.bbc.co.uk/2/low/science/nature/1264205.stm

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

7

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.

Outline

IntroductionDelay 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 destinationData aggregation in two-tier clustering architecture

Outline

IntroductionDelay 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 :2sr

12

Shortest path routing: One dimensional case

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

0rL

0][1][ rerPrP

0][ rerP

01][ 0

rerrE

:Transmission ranger: per-hop progress

)(rE

LH

0r

Two-dimensional case

Per-hop progress

0r

1r

1

2

2r

14/50

Average per-hop progress in 2-D case

220][1][ rePP

2202][ reP

0 0

0

cos][

][

r

ddrP

rE

Average per-hop progress as node density increases

15

Numeric and simulation results

Hop count between fixed S/D distance under various transmission rangeIt shows that our

analysis can provide a better approximation on hop count than .

0r

Hop count simulations

Hop count between various S/D distanceIt shows that our analysis can provide a better approximation on hop count than .

r

Outline

IntroductionDelay 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 (Ts).

Per-hop delay is denoted by d:

2)( sT

dE

sT

s

s

Tds

TdEsd

0

22

12

1)]([)(

19

Random source/dest traffic

Hop count between random S/D pairs

22

2

4)(

22

4/

LLL

P DS

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

Heterogeneous WSN

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

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

21

Random deployment of heterogeneous WSN

N1 = 100N2 = 300L = 1000m

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:

,*

11 LL

N

LL

N

*2

2

2

!

)(),,(

2r

m

em

rrmP

2r

23

2-tier structure

Clusterhead

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

Voronoi diagram

24/50

Distance distribution

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

21

12)( evP

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

1

2)(

vE

Average distance:

25

Average delay in 2-tier WSN

120

2

0 20

),,(

),,()(

2

|)(

rF

T

dvrF

vvP

T

hHdEEDE

s

Ls

Average delay:

Per-hop progress

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.

28

Thanks.

Questions?

29

Random source to central sink node

Laptop computer

30

Incremental aggregation tree

31

Hop count analysis (Key assumptions)