報 告 者:林 文 祺 指導教授:柯 開 維 博士
DESCRIPTION
無線區域網路中自我相似交通流量之 成因與效能評估 The origin and performance impact of self-similar traffic for wireless local area networks. 報 告 者:林 文 祺 指導教授:柯 開 維 博士. Outline. Background of Self-Similarity Properties of WLAN Traffic Estimation of Self-Similar Traffic The Origin of Self-Similarity in WLAN - PowerPoint PPT PresentationTRANSCRIPT
無線區域網路中自我相似交通流量之成因與效能評估
The origin and performance impact of self-similar traffic for wireless local area networks
報 告 者:林 文 祺
指導教授:柯 開 維 博士
Outline Background of Self-Similarity Properties of WLAN Traffic Estimation of Self-Similar Traffic The Origin of Self-Similarity in WLAN Impact of Self-Similar to CSMA/CA
performance Impact of Self-Similar to CSMA/CA
performance with RTS/CTS
Background of Self-Similarity(1/8) Self-Similarity and Fractal
Background of Self-Similarity(2/8)
Statistics of Self-SimilarityDefinition of Self-Similar Stochastic Process:
, 0
H
E x atE x t a
a
2H
Var x atVar x t
a
2
( , ), x
H
R at ast s
axR
H: Hurst parameter or self-similarity parameter
Background of Self-Similarity(3/8)
Self-Similarity of StatisticsDefinition of Self-Similar Stochastic Sequence:( )
( 1)
1, 1,2,3,...
kmmk i
i km m
x x km
3 3 2 3 1 3
3k k k
k
x x xx
( ), 1
2m Var x
Var x Hm
Ex.
m xxR k R k
Background of Self-Similarity(4/8)
Properties of Self-Similarity Long range dependence
Slowly decaying variance
Heavy-tailed distribution
, 0 1C k k as k ,
11 ( ) Pr[ ] ~ , 0F x X x x
x , 當
( )m Var xVar x
m
Background of Self-Similarity(5/8)
Self-Similar Traffic
Background of Self-Similarity(6/8)
X(t) is a Pareto distribution random process with shape parameterαand location parameter k.
1
( )f xk
kx
1 , ; 0k
F x x kx
, 11
E X k
22
2 , 22 1
k
3
2H
Pareto Distribution:
Background of Self-Similarity(7/8) Variance-time Plot
( ), 1,2,3,...m Var x
Var x mm ~
log
log
md Var x
d m
( )log[ ( )] ~ log[ ( )] log( )mVar x Var x m
1 ( / 2)H
R/S Plot
jX =inflow during year j, 1 j N
M N = constant yearly outflow
jL =Reservoir level at end of year j, 1 j N
1
1 N
jj
M N XN
1
j
j kk
L X jM N
11max minj j
j Nj NR N L L
21
1 N
jj
S X M NN
log logR N
a H NS N
Background of Self-Similarity(8/8)
Properties of WLAN Traffic(1/2)
WLAN traffic
Time Unit=1 Sec
Time Unit=0.1 Sec
Time Unit=0.01 Sec
Basic: 1 μS
Aggregation: 1, 0.1, 0.01 Sec
Environment: 7NB
Poisson traffic
Properties of Real Network(2/2)
Time Unit=0.01 Sec
Time Unit=0.1 Sec
Time Unit=1 Sec
Estimation of Self-Similar Traffic(1/2)
• Packets Sequence on WLAN
Estimation of Self-Similar Traffic(2/2)
• Variance Plot & R/S Plot
Single Source without CSMA/CA
The Origin of Self-Similar Traffic (1/3)
The Origin of Self-Similar Traffic(2/3)
• Variance Plot & R/S Plot
The Origin of Self-Similar Traffic(3/3)
• Variance Plot & R/S Plot for WLAN based on single
Poisson Traffic. (Simulated)
Impact of Self-Similar to CSMA/CA performance(1/7)
Maximum throughput The influence of nodes on Self-Similar
Traffic and Poisson Traffic The influence of packet length on Self-
Similar Traffic and Poisson Traffic
Impact of Self-Similar to CSMA/CA performance(2/7)
0
1
2
3
4
5
6
7
8
0 10 20 30 40Node
Thro
ughp
ut(
Mbi
ts)
SS 1000SS 2000SS 250P 1000P 2000P 250
40
50
60
70
80
90
100
0 10 20 30 40Node
Util
izat
ion(
%)
SS 1000SS 2000SS 250P 1000P 2000P 250
Maximum throughput
0
0.001
0.002
0.003
0.004
0.005
0.006
0.007
0 10 20 30 40
Node
Ave
rage
pac
ket d
elay
(se
c)
SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance(3/7)
Maximum throughput
0
50
100
150
200
250
300
350
400
450
0 5 10 15 20 25 30 35 40
Nodes
Num
ber
of c
ollis
ions
SS 1000SS 2000SS 250P 1000P 2000P 250
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
0 20 40 60 80Load(%)
Thro
ughp
utM
bits
()
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P, N=20 0
10
20
30
40
50
60
70
80
0 20 40 60 80Load(%)
Util
izat
ion(
%)
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P,N=20
Impact of Self-Similar to CSMA/CA performance(4/7)
The influence of nodes on Self-Similar Traffic and Poisson Traffic
0
50
100
150
200
250
0 20 40 60 80Load(%)
Num
ber o
f col
lisio
ns
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P, N=20
0
0.0005
0.001
0.0015
0.002
0.0025
0 20 40 60 80Load(%)
Aver
age
pack
et d
elay
(se
c)
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P, N=20
Impact of Self-Similar to CSMA/CA performance(5/7)
The influence of nodes on Self-Similar Traffic and Poisson Traffic
40
50
60
70
80
90
100
20 40 60 80 100Load(%)
Util
lizat
ion(
%)
SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance(6/7)
The influence of packet length on Self-Similar Traffic and Poisson Traffic
2
2.5
3
3.5
4
4.5
5
5.5
6
6.5
7
20 40 60 80 100
Load(%)
Thro
ughp
ut(M
bits
)
SS 1000SS 2000SS 250P 1000P 2000P 250
0
50
100
150
200
250
20 40 60 80 100Load(%)
Num
ber o
f col
lisio
ns
SS 1000SS 2000SS 250P 1000P 2000P 250
0
0.0002
0.0004
0.0006
0.0008
0.001
0.0012
20 40 60 80 100Load(%)
Aver
age
pack
et d
elay
(se
c)SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance(7/7)
• The influence of packet length on Self-Similar Traffic and Poission Traffic
Maximum throughput The influence of nodes on Self-Similar
Traffic and Poisson Traffic The influence of packet length on Self-
Similar Traffic and Poisson Traffic
Impact of Self-Similar to CSMA/CA performance with RTS/CTS (1/4)
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
0 5 10 15 20 25 30 35 40Node
Thro
ughp
ut(
Mbi
ts)
SS 1000SS 2000SS 250P 1000P 2000P 250
50
55
60
65
70
75
80
85
0 5 10 15 20 25 30 35 40Node
Util
izat
ion(
%)
SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance with RTS/CTS (2/4)
Maximum throughput
1
1.5
2
2.5
3
3.5
0 20 40 60 80Load(%)
Thro
ughput
Mbits
()
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P, N=20
0
10
20
30
40
50
60
70
80
0 20 40 60 80Load(%)
Util
izat
ion(
%)
SS, N=5SS, N=10SS, N=20P, N=5P, N=10P, N=20
Impact of Self-Similar to CSMA/CA performance with RTS/CTS (3/4)
The influence of nodes on Self-Similar Traffic and Poisson Traffic
1
1.5
2
2.5
3
3.5
4
4.5
5
5.5
6
20 40 60 80 100Load(%)
Thro
ughp
ut(
Mbi
ts)
SS 1000SS 2000SS 250P 1000P 2000P 250
35
40
45
50
55
60
65
70
75
80
85
20 40 60 80 100Load(%)
Util
izat
ion(
%)
SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance with RTS/CTS (4/4)
The influence of packet length on Self-Similar Traffic and Poisson Traffic
Conclusion WLAN Traffic is Self-Similar regular & Single) WLAN Throughput at node=5 Max WLAN Throughput at node<5 Poisson>SS WLAN Throughput at node>5 Poisson<SS Impact of Packet Length RTS/CTS not influence the characteristic of Poisson and
Self-Similarity
Thanks for your attendance
Impact of Self-Similar to CSMA/CA performance The Number of Nodes increment form 1 to 5
0
1
2
3
4
5
6
7
8
1 2 3 4 5Node
Thro
ughput(
Mbits)
SS 1000SS 2000SS 250P 1000P 2000P 250
50
55
60
65
70
75
80
85
90
1 2 3 4 5Node
Utiliz
ation(%
)
SS 1000SS 2000SS 250P 1000P 2000P 250
Impact of Self-Similar to CSMA/CA performance The Number of Nodes increment form 1 to 5
0
20
40
60
80
100
120
140
1 2 3 4 5
Node
Num
ber
of
colli
sion(%
)
SS 1000SS 2000SS 250P 1000P 2000P 250
0
0.05
0.1
0.15
0.2
0.25
1 2 3 4 5
Node
Ave
rage
pac
ket d
elay
(se
c)SS 1000SS 2000SS 250P 1000P 2000P 250