EasiTPQQoS BasedTopology Control in Wireless Sensor Network

foamyflumpMobile - Wireless

Nov 21, 2013 (3 years and 8 months ago)

68 views

EasiTPQ

Qo匠Sa獥s

Topology
C
ontrol
in
W
ireless
S
ensor
N
etwork



L
IU

Wei

1
Institute of
Computing

Technology, Chinese Academy of
Sciences,
Beijing

100080

2
Graduate

University of Chinese
Academy of
Sciences, Beijing

100049

CUI Li


Institute of Computing
Tech
nology, Chinese Academy of
Sciences, Beijing 100080



LI Tian Pu

I
1
nstitute of Computing
Technol
I
ogy, Chinese Academy
of Sciences, Beijing 100080

2
Graduate

University of Chinese
Academy of
Sciences, Beijing

100049



Abstract

With

the
rapid
development of

wireless sensor network
,
the requirements for
quality of service are

g
rowing,

particularly in
applications

where

real time imaging,
video or audio
communication
s

are involved
.
The system
needs to
meet optimized

energy consumption design
criterion
a
nd

to s
atisfy
certain

QoS requirement
s

at the
same tim
e from long view
.


Most of the existing works
deal with

resource allocation (e.g., scheduling or
buffering) or

routing
strategy

to
achieve
QoS
. In

this
paper,

we propose to extend
QoS

support to
the
topology
c
ontrol layer

by
introducing

a number of
active nodes
distributed in a gradient fashion based on the
ir

logical
distance
s

to the sink node
.

Present

a

novel
topology
control

algorithm,

namely

EasiTPQ (
Easy QoS based
topology control)

for QoS
improvement

in wi
reless
sensor networks.

Simulation results show that data loss
rate and latency are
average
ly

improved by
6
0% and
5
5%
,

respectively
.


Key words
:

gradient

change

density; QoS
; state transmission


1
.
Introduction


A wireless sensor network is

composed of

a large
number
of tiny

and inexpensive

sensor nodes which can
comput
e and communicate by wireless [
1]
.

S
ensor nodes
can be

deployed

densely

on a large scale

and

are
hard to
be re
-
collected and recharged.
Thus

the
level of
energy
consumption is an importan
t
design criterion for

sensor
networks, since it is directly related to the network
lifetime
.

Topology control and management is an
effective way to control the
power
level

of the

network.
H
ere

how

sensor nodes cooperate with each other to

create
an

effici
ent and
connect
ed network to upper layer
is the
major problem in
t
opology control

process.


Although most of
the
application
s

are
dealing with

small data flow obey
ing

the principle of
best effort
, the
high
-
speed data flow
, however,

have to be

sustained
fro
m long view
.

T
his trend looks more apparent with the
introduction of
hybrid sensor
network
s

[
2
].

In this
work
,
research

was focused

on
a
QoS
-
based topology control
algorithm

which can

reduce packet loss rate and latency
in networks where

some
sensor nodes
generate
high
-
speed data to sink simultaneously.
Energy saving
scheme was also
designed. Nodes

which

have little

contribution

to

QoS
guarantee

were
schedule
d to

sleep

to

prolong
the
life time

of the network
.

The rest of this paper is organized as follows:
Section
II summarize
s

the
related

work
s
.
Section III describes
our QoS
-
based topology control algorithm
and analyzes

some crucial parameter selection
s

from
simulation.
Finally
, Section IV
concludes

the paper.


2
.
Related

Works


In topology control, both
powering off redundant
nodes and lowering radio power while keeping the
This research was supported by NSFC key project under Grant No. 60533110
;

NSFC

general pr
oject under Grant No. 60572060;

CAS

Project of 100 Talents

.


network connected can contribute to power
-
saving.
However
, powering off
unnecessary
nodes can

achieve
better results

[
3
-
5
]
.

Since sensor nodes

ha
ve

limited

capability of
computing and
limited

memory space to storage data
, the

QoS
solution

for sensor networks
can not simply follow
those for

br
oad
-
band

network
s
.
Moreover,

sensor
network is highly
oriented

to application
s
,

so

that

a
“fits
-
all” QoS

solution
is unlikely existing.

Sequential
Assignment Routing (SAR) [
6
] is the first
protocol

for
sensor networks

that includes a notion of
QoS. Assuming

multiple paths to the sink node, each
sensor uses a SAR

algorithm for path selection. It takes
into account the energy

and QoS factors on each pa
th
.

SPEED [
7
] is similar
to

SAR
.

T
he
protocol

requires
each node to maintain its

neighbor’s

information
.

In
addition, SPEED strives to ensure a certain speed for
each

packet delivery so that each application can
estimate the end
-
to
-
end delay for the packet
s by
considering the distance to the

sink and the speed of the
packet delivery before making the

admission decision
.

For r
eal
-
time traffic generated by imaging

or video
sensors
,

t
he protocol

presented in [8]

finds a least cost

and energy efficient path tha
t meets certain end
-
to
-
end
delay

requirement.

More recently,
hybrid sensor
network

attracts

more
attention.
A three
-
tired

architect
ure presented
in

[
2
] are
composed of
normal

data collection nodes, data relay
nodes and data relay gateway.
A path composed
of high
capacity
and little latency
nodes can be selected


3
.

QoS
-
Based Topology Control (EasiTPQ)


3.1
.

The Feasibility and
necessity

of EasiTPQ


Existing

research on QoS problem in wireless sensor
network is focused on how to find
a
proper data path

in
pre
-
existing network topology

on

the hypothesis
that
this
path
exist
s. A
ctually path
satisf
ying

all the

QoS

requirements
does not

always
exist


especially
topology
control
scheme is

adopted
.

By

analyzi
ng

the architecture of WSN,
we conclude

that
each

node

in
network has

different functional
ities

in
transmitting data
.
T
he data transmission
mode
is

usually

in
a
multi
-
to
-
one
fashion where m
ore relay data flow
s

are

transmitted
by

node
s

closer to the

sink.
Hence for
those

interior

nodes,
the tasks

of
transmi
tti
ng data may
be quite heavy with more

relay
data than

local generated
data
.
Meanwhile

the exterior nodes trans
mit
local

collected data mostly.

T
he most exterior nodes have
only
got
local

collect
ed

data

to transmit.



3.2
.

Gradient

Topology Structure


3.2.1

EasiTPQ Design


Based on the above analysis, t
he proposed
QoS
-
based
topology control
algorithm
, namely
EasiTPQ
,

is
designed
by reducing
exterior
redundant

nodes
and

increasing

interior
active
nodes
simultaneous
ly

to create
a gradient topology
structure

ai
ming
to
meet

QoS

requirements
,
e.g.

the data loss rate and
latency
.

Fig.
1

illustrates

the nodes category
created by
our EasiTPQ
algorithm
.


Fig.
1

Architecture of Ea
siTPQ

There
are

three types of nodes in
EasiTPQ
:

(a) NA
(Not Active)
nodes which are nodes
in

sleep mode, (b)
CA (Connectivity Active) nodes which are scheduled to
work for network
connecti
on

and (c) QA (QoS Active)
nodes which
are scheduled to work for extra sustain
to
meet

for

QoS requirement
s
. The main
design
criterion

of
EasiTPQ

is to add QA

nodes as

few as possible

to satisfy
certain

QoS requirement
s
.

T
hree

definitions

are introduced as
below:


Definition 1: Node

s active power value
A
P
:

A
P

denote
s

the value or capacity to become an
active node
.

T
P
N
E
e
A









1


Where

e

is

node

s remain
ing power
, and
E

is

node

s
initial

power.
T
N

Neighbor Total


is the number of

node’s

neighbor it can
communicate

whiles

in non
-
sleep
state.
T
N

reflect
s

the importan
ce of node

in

the

netwo
rk and can
be used to
differentiate

crucial nodes
from other normal
ones
.


and


are

the
weight

value

between
this
two parameters,

and can be
flexibl
y
configured according specific application.

Definition 2:
Th
e sorting function
p
A

The

function

)
(
N


is

the
value
order
of local node

s
p
A
among all
p
A
values it can hear.




}
1
),
(
:
{
)
(
_
N
i
i
A
A
N
N
p
local
p






(2)

local
p
A
_

is

the
p
A

value of local node, and
)
(
i
A
p

is

other node

s
p
A

value
th
is
local node can
hear

at

its R_Test stage. So
definition 2 reflects

a
p
A
sorting relation between local nodes and
its
neighbors.

Definition 3:
G
radient
descending
function
)
(
M


)
(
M


is

the

relation

between node

s
active

neighbor number and

its

hop count

to sink.


)
1
(
)
(
)
(
1





i
M
n
h
M
i




3
-
1





1
)
(
)
(
2





i
M
n
h
M
i



3
-
2





In this paper, we focus our attention mainly on two
types of function
, including a

liner

descending

function
and
a
square root
descending

function described in
formula (3
-
1) and (3
-
2) res
pectively. The
reason

of
selecting
these

two types of function is
that

these two
simple functions can be used as examples to

evaluate

the
e
ffect of
node

s
gradient distribution s
cheme
s

to
the
overall QoS

performance

of
the
network.
M

is

the initial
number
of neighbors and
is often
set by

the
value of the
diameter of network.

i
i
s
the hop count between
each
node to
the
sink node and
the

initial value is set to 1.
Thus

)
(
n
h
i

represents

the
active
neighbor

number of
local node which have a ho
p
count

i
to sink node.


3.2.
2

EasiTPQ

implementation


In
EasiTPQ
, nodes are in one of
the
four states:

R_Test
,
R_Passive
,
R_Active

and
R_Sleep
. Fig.
2

shows a state
transition diagram.



Fig.
2

Diagram of
EasiTPQ state transitions

(1)
Initially,
only sink node is
in

the active state and
other nodes are in
R_Passive

state

with

their radio on to
overhear all packets transmitted by their active neighbors.
No data packets are forwarded in this state since this is a
listen
-
only state.

(2) Sink node sen
d
s

Topo_Initial command by
broadcasting which inform
s

nodes in network to
enter
the process of
EasiTPQ

algorithm
.

(3)
Nodes received Topo_Initial command
enters

R_Test
state from
R_Passive
state, and start to exchange data
and routing control messages.

(
4) Nodes in
R_Test
state start to broadcast message
including
P
A

value
and,
at the same time
, to

set up a
timer
T
t
. When
T
t

expires, the node enters the
R_Active
state.
B
efore
T
t

expires
,

if

the
node finds

that the

average
data loss ra
te
L
P
is higher than the average loss before
entering in the
R_Test

state, then the node moves into the
R_Passive

state.
0
N

is
the
active neighbor count need
to
be
decided

with

a

minimal value

of

min
N
.
min
N

is
set in view of the basic network connectivity
target [
9
].
From above
information

we know that
0
N

is a crucial
parameter in
EasiTPQ

and this

value has
a gradient
property which changes

according to hop c
ount
from

this
node

to

the
sink.
0
P

is the average data loss before
entering in
R_Test

state,
which
is
the
influence

of

data
loss rate when this node enter
the
data
transmission

process
.

(5)
Nodes in
R_Passive
state set up a timer
T
p
.

When
T
p

expires
, the

node enters the
R_Sleep
state.
B
efore
T
p

expires
,
if
the number of neighbors is below
min
N

or
the data loss rate
L
P

higher

than
T
P
,
the node

makes a
transition to the
R_Test

state to transmit data.
T
P
denotes
the
minimal

data loss rate application required, which is
a parameter
related to the

QoS property.

(6) Nodes in
R_Active
state transfer to
R_Test

state after
a period of time
T
a

to detect whether it s
hould become an
inactive

node. This detection process
may

balance the
power consumption among nodes, and
may

avoid
the

crucial node
s

consume all its power
to make

other nodes
logically
br
oken

down

(
e.g.

could not
transmit

data
continually
). A node
which

en
ters the
R_Sleep
state turns
the radio
off, sets

a timer
Ts
, and goes to sleep. When
Ts

expires, the node moves into
R_Passive
state.

After all above procedures, the active nodes form
a
network architecture
consist
ing

of three types of nodes.
The neighbor
number

min
N
ensure
s

the basic network
connectivity [
9
], while other
active

nodes or QA
nodes
ensure

some QoS
performance

in network.


3.
3
.

Performance evaluation and
p
arameter
analysis


In th
e

simulation

part
,
we use NS2 simulation tools
and
choose parameter
s as

below:
6
.
0



4
.
0


,
%
15

T
P

Tp
=2min

Ts
=2min

Tt
=4min

Ta
=4min

min
N
=2,
and
node number is 120.

N
etwork
performance
s were examined

according to different
)
(
N


scheme
s as shown in (3
-
1) and (3
-
2),
respectively
.

The simulation results of

data loss rate and data
latency
are shown
in Fig.
3
.
PL1

and
LA1

is the initial
network data loss rate and data latency whose topology
is obtained by
ASCENT [
9
].

PL2

a
nd

LA2

is the network
data loss rate and data latency whose topology is
obtained by EasiTPQ using formula (3
-
1)
(in a
liner
descend
ing scheme
)
. And
PL3

and

LA3

is the network
data loss rate and data latency whose topology is
obtained by EasiTPQ using formu
la (3
-
2)

(in a
square
root
descend
ing
scheme
)
.

0
200
400
600
800
1000
1200
4
8
12
16
20
24
data rate(kbps)
Latency(ms)
LA1
LA2
LA3

0
5
10
15
20
25
30
4
8
12
16
20
24
data rate(kbps)
Packet loss rate(%)
PL1
PL2
PL3

Fig.
3

Packet

loss rate
and Latency




From Fig.
3
, we can see that


(a)

The
performance

in PL2 is
improved by 50%


Averagely

than that of

PL1,
whi
ch can verify the
efficiency of a
dding

QA nodes?


(b)
The performance of PL3
decrease
fast

when data rate
exceed
certain

value. This

may

because
those more extra
active QA nodes

in exterior circle may cause more
channel competition. When data rate

exceeds some
threshold, the
negative

effect of ch
annel competition will
suppress

the QoS contribution by
these extra active QA
nodes
.

(c) PL3 ha
s

a
n
improvement by
70%
in
the
data rate

range of

0
-
20Kbps
.M
any

real time imaging and audio
communication
s are

in this range.
O
ther

application
s

with

high
er data rate than the

threshold
may

choose PL2
scheme

for better QoS performance
.

(
d
) The
latency from
node to sink
of

LA3

is

always less
than that of LA1

with an
improve
ment

by
4
0%

in
average.

This may because t
hose packages successfully
transmitted have more easy access to MAC channel, so
less channel access latency can be acquired.

(
e
) LA2
has

a very good latency performance
.
The
latency of

LA
2 is

less than that of LA1

with an
improvement by averagely 70%

in
th
e
data rate
range of

0
-
20Kbps
.
This may because that LA2 scheme both have
a less channel access latency and have less extra hop
latency compare to LA3 scheme in
that data rate

range
.
But when the data rate
exceeds

some threshold

the
MAC
latency might be ap
parent compare
d

to other
factors
which are

the same as LA1, if considering the extra hop
latency the larger latency may
occur
.


4
.

Conclusions


The
EasiTPQ

algorithm
proposed
in this paper is a
QoS based topology control algorithm.
EasiTPQ

make
s

use of the

fact that e
ach

node in network has different
function
alities

in data
transmission
, e.g.,
some nodes
bear more data relay task
s

whil
st

some
other
node
s

only
transmits

data generated

by itself
.
EasiTPQ

schedule
more nodes active if these nodes are
responsib
le more
to

relay


data task
s
.
S
imulation

result
s

show that
EasiTPQ

enables

improved

data loss rate and delay
performance

by
60% and 55%
respectively

compare
d

with

another existing

topology control algo
rithm

such as
ASCENT
.
EasiTPQ

is more
useful

if the da
ta rate range
is
within

a certain

range
.

The QoS problem is highly
oriented

to application in wireless sensor network
.

F
urther
experiment
al

or simulation
works need to be
carried out to study the
conc
rete parameter
settings for
particular applications.


Re
ferences


[1]
崔莉
,
鞠海玲
,
苗勇
,
李天璞
,
刘巍
.
无线传感器网络研究进

.
计算机研究与发展
,2005,42(1):163
-
174

[
2
]
O.Kasten. Energy

consumption. ETH
-
Zurich, Swiss Federal
Institute of Technology. Available at
http://www.inf.ethz.ch/˜kasten/research/bathtub/energy_consu
mption.html%, 2001.

[
3
]

Z
afer Sahinoglu and Philip Orlik

“transmit power

adjustment,” in Proc. IEEE INFOCOM, 2000, pp.

404

413.
[Online]. Available: citeseer.

nj.nec.com

/ramanathan00topology.html

[
4
]
G.C˘alinescu, I.M˘andoiu, and A.
Zelikovsky, “Symmetric
connectivity with minimum

power consumption in radio
networks,” in 2nd IFIP International Conference on Theoretical
Computer Science (TCS 2002). Kluwer Academic Publishers,
2002, pp. 119

130.

[
5
]
M.Stemm and R.H.
Katz. Measuring and reducing energy
consumption of ne
twork interfaces

in hand
-
held devices. IEICE
Transactions on Communications, E80
-
B(8):1125

1131, Aug.
1997.

[
6
]
B.Chen, K.Jamieson, H.Balakrishnan, and R.
Morris. Span:
An energy
-
efficient coordination algorithm for topology
maintenance in ad hoc wireless net
works. ACM Wir
eless
Networks, 8(5), September 2002.

[
7
]
C. Schurgers, V. Tsiatsis, and M.

Srivastava.

STEM:
Topology management for energy efficientsensor networks. In
IEEE Aerospace Conference,pages 78

89, March, 2002.

[
8
]
K. Sohrabi, J. Gao, V. Ailawadhi and G. Pottie
, “Protocols
for Self
-
Organization of a Wireless Sensor Network, ” IEEE
Personal Communications, pp. 16
-
27, October 2000.

[9]
A.Cerpa and

D.
Estrin. ASCENT: Adaptive
selfconfiguring sensor network topologies. In Twenty
First International Annual Joint Confe
rence of the IEEE
Computer and Communications Societies (INFOCOM),
June 2002