LICENTIATE T H E S I S
Luleå University of Technology
Department of Computer Science and Electrical Engineering Division of Systems and Interaction
2006:23|: 02-757|: -c -- 06 ⁄23 --
2006:23
On the Performance of Transmitting IP Traffic over a Shared Cellular Radio Channel
with a Central Scheduling Mechanism
Mats Folke
A BSTRACT
C ONTENTS
T
HESISI
NTRODUCTIONP
APERP
APERP
APERP
APERR
EFERENCESA CKNOWLEDGEMENTS
T HESIS I NTRODUCTION
1 Background
1.1 Congestion control and avoidance
T
HESISI
NTRODUCTION1.2 History of cellular networks
1. B
ACKGROUNDT
HESISI
NTRODUCTION1.3 Scheduling in cellular networks
1.4 Scope of the thesis
2. M
ETHODOLOGY–
ABOUT SIMULATION2 Methodology – about simulation
T
HESISI
NTRODUCTION3 Future work
4 Short presentation of the papers
4. S
HORT PRESENTATION OF THE PAPERS4.1 On the TCP Minimum Retransmission Timeout in a High-speed Cellular Network
4.1.1 My contribution
4.1.2 Impact
4.2 On the Influence of User Behaviour and Admission Control
on System Performance in HS-DSCH
T
HESISI
NTRODUCTION4.2.1 My contribution
4.2.2 Impact
4.3 Scheduling Support for Mixed Conversational and Background Traffic over HSDPA
4.3.1 My contribution
4.3.2 Impact
4.4 An NS Module for Simulation of HSDPA
4.4.1 My contribution
4.4.2 Impact
P APER
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
mats.folke@ltu.se sara.landstrom@ltu.se ulf.bodin@ltu.se
1 Introduction
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
2 TCP fundamentals
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
3 Method
3.1 Simulation Environment
Mobile nodes 25ms/5Gbps
Traffic sources
3.2 Evaluation metrics
x
iO
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
f (x
1, x
2, · · · , x
n) =
n
i=1
x
i 2n
ni=1
x
2i4 Results
5 Discussion
0 0.05 0.1 0.15 0.2 0.25
0 50 100 150
Share of spurious timeouts
Number of users 0.0s
0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
0 0.05 0.1 0.15 0.2 0.25
0 50 100 150
Share of spurious timeouts
Number of users 0.0s
0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
0.5 0.55 0.6 0.65 0.7 0.75 0.8
0 50 100 150
Fairness in goodput
Number of users
0.0s 0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
0.5 0.55 0.6 0.65 0.7 0.75 0.8
0 50 100 150
Fairness in goodput
Number of users
0.0s 0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
1e+06 2e+06 3e+06 4e+06 5e+06 6e+06 7e+06
0 50 100 150
Throughput [bits/s]
Number of users 0.0s
0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
1e+06 2e+06 3e+06 4e+06 5e+06 6e+06 7e+06
0 50 100 150
Throughput [bits/s]
Number of users 0.0s
0.1s 0.2s 0.3s 0.4s 0.5s 1.0s
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
6 Conclusions
O
N THETCP M
INIMUMR
ETRANSMISSIONT
IMEOUT...
P APER
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
mats.folke@ltu.se ulf.bodin@ltu.se
1 Introduction
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
2 Method
2.1 Simulation Environment
2.1.1 Radio model
2.1.2 User mobility
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
2.1.3 Application and load model
2.2 Evaluation metrics
f (x) =
⎧ ⎪
⎨
⎪ ⎩
0 x < 100000
x−100000
300000
100000 ≤ x < 400000
1 x ≥ 400000
3 Results
3.1 Without user behaviour modelled
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
10 15 20 25 30 35 40 15 10
25 20 35 30 40 1e+07 1.2e+07 1.4e+07 1.6e+07
Total goodput
Rate Max users
Total goodput
15 10 25 20 35 30 40 10 15 20
25 30 35 40 1000
2000 3000 4000 Total satisfaction
Rate Max users
Total satisfaction
10 15 20 25 30 35 40 15 10
25 20 35 30 40 0 100 200 300 Active users
Rate Max users
Active users
3.2 With user behaviour modelled
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
10 15 20 25 30 35 40 100 150 200
250 300 350400 1e+07
1.2e+07 1.4e+07 1.6e+07
Total goodput
Rate Min thp.
Total goodput
15 10 25 20 35 30 100 40 200150 300250 400350 1000 2000 3000 4000 Total satisfaction
Rate Min thp.
Total satisfaction
10 15 20 25 30 35 40 100 150 200
250 300 350400 0
100 200 300 Active users
Rate Min thp.
Active users
10 15 20 25 30 35 40 15 10
25 20 35 30 40 1e+07 1.5e+07 2e+07 2.5e+07
Total goodput
Rate Max users
Total goodput
15 10 25 20 35 30 40 10 15 20
25 30 35 40 1000
2000 3000 4000 Total satisfaction
Rate Max users
Total satisfaction
10 15 20 25 30 35 40 15 10
25 20 35 30 40 0 100 200 300 Active users
Rate Max users
Active users
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
4 Conclusions
10 15 20 25 30 35 40 100 150 200
250 300 350400 1e+07
1.5e+07 2e+07 2.5e+07
Total goodput
Rate Min thp.
Total goodput
15 10 25 20 35 30 100 40 200150 300250 400350 1000 2000 3000 4000 Total satisfaction
Rate Min thp.
Total satisfaction
10 15 20 25 30 35 40 100 150 200
250 300 350400 0
100 200 300 Active users
Rate Min thp.
Active users
O
N THEI
NFLUENCE OFU
SERB
EHAVIOUR ANDA
DMISSIONC
ONTROL...
P APER
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
mats.folke@ltu.se sara.landstrom@ltu.se
ulf.bodin@ltu.se stefan.wanstedt@ericsson.com
1 Introduction
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
PF
MR MR
MR
minMR
delay2 Methodology
2.1 Metrics
2.2 HS-DSCH model
2.3 Application model
2.3.1 VoIP traffic
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
2.3.2 Web traffic
2.3.3 Load model
2.4 Schedulers
PF
i∗ = arg max
i
r
iμ
iμ
ii r
iMR
i∗ = arg max
i
{r
i(1 + βe
−β(μi−μmin))}.
β MR
minμ
minPF MR
minMR
delay3 Results
3.1 Cell throughput
MR
minMR
delayMR PF MR PF
MR
delayMR PF MR
minMR
delayMR
minMR PF
MR PF
3.2 User perceived quality
MR
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
600000 800000 1e+06 1.2e+06 1.4e+06 1.6e+06 1.8e+06 2e+06 2.2e+06
5 10 15 20 25 30 35
Cell throuhgput [bits/s]
Number of users 40 ms
80 ms 120 ms 160 ms
MR
600000 800000 1e+06 1.2e+06 1.4e+06 1.6e+06 1.8e+06 2e+06 2.2e+06
5 10 15 20 25 30 35
Cell throuhgput [bits/s]
Number of users 40 ms
80 ms 120 ms 160 ms
PF
600000 800000 1e+06 1.2e+06 1.4e+06 1.6e+06 1.8e+06 2e+06 2.2e+06
5 10 15 20 25 30 35
Cell throuhgput [bits/s]
Number of users 40 ms
80 ms 120 ms 160 ms
MRmin
600000 800000 1e+06 1.2e+06 1.4e+06 1.6e+06 1.8e+06 2e+06 2.2e+06
5 10 15 20 25 30 35
Cell throuhgput [bits/s]
Number of users 40 ms
80 ms 120 ms 160 ms
MRdelay
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
MR PF
PF PF
MR
delayMR
delayMR
delayMR
minMR PF
MR
minMR
MR
3.3 Web users with throughputs lower than 15 kbps
MR
minPF
MR MR
delay4 Discussion and conclusions
PF MR
MR
minMR
delay0 0.2 0.4 0.6 0.8 1
5 10 15 20 25 30 35
Ratio of satisfied users
Number of users
VoIP traffic Web traffic
MR
0 0.2 0.4 0.6 0.8 1
5 10 15 20 25 30 35
Ratio of satisfied users
Number of users VoIP traffic, 40 ms
PF
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
0 0.2 0.4 0.6 0.8 1
5 10 15 20 25 30 35
Ratio of satisfied users
Number of users VoIP traffic, 40 ms
MRmin
0 0.2 0.4 0.6 0.8 1
5 10 15 20 25 30 35
Ratio of satisfied users
Number of users Web traffic 40 ms
Web traffic 80 ms Web traffic 120 ms Web traffic 160 ms VoIP 160 ms VoIP 120 ms VoIP 80 ms VoIP 40 ms
MRdelay
MR
minMR
delayMR
delayMR
minPF
PF PF
S
CHEDULINGS
UPPORT FORM
IXEDC
ONVERSATIONAL...
P APER
A
NNS M
ODULE FORS
IMULATION OFHSDPA
mats.folke@ltu.se sara.landstrom@ltu.se
1 Introduction
1.1 Simulation of wireless systems
A
NNS M
ODULE FORS
IMULATION OFHSDPA
2 HSDPA
2.1 Media Access Control
A
NNS M
ODULE FORS
IMULATION OFHSDPA
Core Network
RNC RLC MAC
Node B MAC−hs
PHY
Node B MAC−hs
PHY
UE UE UE UE UE
2.1.1 Fast Link Adaptation
2.1.2 Transport block selection process
R(t) = (0.2 ∗ x(t)
3+ B) log
2(1 + 10
x(t)/10)
R(t) x(t)
B
k
tA
NNS M
ODULE FORS
IMULATION OFHSDPA
k
t2.1.3 Fast Hybrid ARQ
2.1.4 Fast Scheduling
i
r
ii r
ir
mini d
id
prioi
S
ii S
ii∗
A
NNS M
ODULE FORS
IMULATION OFHSDPA
i∗ = arg max
i
{S
i}.
i∗ = arg max
i
{r
i}.
i∗ = arg max
i
r
ir
i.
r
ir
ir
ir
ii∗ = arg max
i
S
iS
i.
i∗ = arg max
i
{r
i(1 + βe
−β(ri−rmin))}.
r
minr
ir
minβ
i∗ =
⎧ ⎨
⎩ arg max
i
{r
i} d
i< d
prio, arg max
i
{d
i+ R
max}
R
maxd
id
prio2.2 Radio Link Control
2.3 Cell selection
A
NNS M
ODULE FORS
IMULATION OFHSDPA
2.3.1 Admission control
3 Propagation environment
•
•
•
3.1 Path loss, shadowing, and multi-path fading
P L(d)[dB] = P L(d
0) + 10n log
d d
0+ X
σn n X
σσ d d
03.2 Interference
A
NNS M
ODULE FORS
IMULATION OFHSDPA
0 50 100 150 200 250 300 350
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Autocorrelation profile for shadowing
Distance moved between updates of the shadowing values [metres]
Autocorrelation
Correlation Distance 40 metres 1/e
α
α = 0.6 α = 0.9