Professional Documents
Culture Documents
HSUPA
HSUPA
which benefits both end users and providers. The request for
improved networks supporting data communications was very
high, which led to Universal Mobile Telecommunication
System (UMTS).
Although UMTS offered bit rates up to 384 kbps [1], they
were not high enough for customers needs. In order to offer
improved bandwidth to end-user, improved network capacity
to the operator and improved interactivity for multimedia
applications, 3rd Generation Partnership Project (3GPP)
published Releases 5 and 6, also known as High Speed Packet
Access (HSPA) [2]. First was Release 5 in March 2002 for
Downlink (DL), enabling peak rates initially up to 1.8 Mbps,
increasing to 3.6 Mbps and 7.2 Mbps, and probably beyond
10 Mbps. In December 2004 it was Release 6 for Uplink
(UL), also known as High Speed Uplink Packet Access
(HSUPA). The first HSUPA networks were deployed during
2007, achieving peak rates around 1-2 Mbps, and in a near
future around 3-4 Mbps.
HSUPA presents some new features like a new physical
channel, the Enhanced Dedicated CHannel (E-DCH), fast
physical layer Hybrid Automatic Repeat Request (HARQ),
fast Node B based scheduling and a shorter UL Transmission
Time Interval (TTI) length of 2 ms, while higher order
modulation showed no potential gains. However, no potential
gains from using higher order modulation in the UL direction
wer reported.
The main objective is studying the impact of HSUPA
implementation on UMTS capacity and cell coverage. In
order to do so a theoretical model will be developed based in
Admission Control algorithm for a single sector, and will
present results for the number of users served, the achievable
throughput, cell radius and traffic processed. Later on this
work, the comparison between theoretical HSUPA results and
OPNET Modeller simulation results will be done. However,
only Release 99 will be simulated, as OPNET does not
include HSUPA implementation.
Section II describes the theoretical model and assesses the
scenarios and applications. It is also done the analysis of
results. In Section III, the used simulator, corresponding
scenarios, and obtained results are presented. Also, a
comparison between the obtained results with the theoretical
2
model, and with the simulator is done. In section VI some
conclusions are drawn.
II. CAPACITY AND COVERAGE MODEL
In order to study the impact of HSUPA implementation on
UMTS capacity and cell coverage a simple theoretical model
was developed. It is mainly based on the Admission Control
algorithm for a single sector, and it presents results about the
networks behaviour, like: number of served users; available
throughput per user; maximum cell radius.
QoS class
Service
Streaming
Streaming
Interactive
FTP
Interactive
HTTP
Background
Background
MMS
A. Model
HSUPAs snapshot analysis it is an improvement from the
work by developed [3] and [4], which assesses a mono-service
scenario. This work presents the following features: multiservice (CS voice and video-telephony, and PS streaming,
FTP, HTTP, e-mail and MMS) and dynamic carrier (in order
to support, at the same time, both R99 and HSUPA clients)
for the UL. As mentioned before, it is a snapshot analysis,
which means evaluating the behaviour of the network in a
certain instant of time
N u : number of users;
MT
(1)
being:
R : cell radius.
And the total path loss, according to the propagation
model, is:
(2)
LP[dB] = L0[dB] + Ltt[dB] + Ltm[dB] ,
where:
Ltt : losses due to propagation over roof tops;
Ltm : losses due to diffraction from the roof to the MT.
Manipulating (1) and (2), the cell radius can be determined
by:
d[km] = 10
(3)
(4)
i =1
being:
i : number of calls per hour for service i;
N i : number of served users for service i;
Vi : mean file volume for service i;
N S : number of different services.
The HSUPA throughput is based on the UL load factor [2].
Knowing the available load factor in UL, then it is possible
assessing the number of HSUPA users, and their throughput.
The total UL throughput is given by:
UL[kbps] = Nu R 99 CS 12.2 + N uR 99 PS 64 + N uHSUPA a [kbps]
(5)
B. Scenarios
It is considered all users are at a maximum distance. In
3
Table II, the percentage of users for each service profile is
presented. These values determine the number of users that is
doing each service.
80
70
T ABLE II
Users [%]
Voice Centric
60
50
Data Centric
R99 PS
MMS
40
Voice
48.6
22.3
Video Telephony
0.2
0.3
Streaming
7.1
10.6
FTP
16.9
25.5
10
0
HTTP
11.8
17.7
10.5
15.9
MMS
4.9
7.7
64
1450
R99 PS
64
TABLE IV
DEFAULT VALUES USED IN LINK BUDGET.
Parameters
Value
24.0
UL frequency [MHz]
[1920; 1980]
17
Release 99
1.5
HSUPA
Not considered
[0.391; 0.65]
20
30
40
50
60
70
80
90
100 110
120
130
140
150
12.2
HSUPA
10
Number of users
Throughput [Kbit/s]
Video-Telephony
VT
Voz
TABLE III
PS SERVICES TARGET THROUGHPUT.
Voice
Strm
20
Service
E-mail
FTP
30
Voice
Data
5.0
UL Load Factor
0.9
2.0
Diversity Gain
3.0
C. Results
In this section, results are presented. The objective is to
understand network behaviour for an increasing number of
users, when each of them is trying to make a different service.
This assessment is made for the pedestrian environment.
Fig. 1 presents the user distribution for the Voice Centric
(VC) profile. Data Centric (DC) also has a similar behaviour.
However, in order to respect the target throughput, the latter
one serves all clients until they are 15, while the former one
until they are 21. Another difference is the fact that, while for
the VC profile there are HSUPA users served until the
4
250
Ped-Veh (R99)
Ped-Ind (R99)
Light
Heavy
Linear (Heavy)
200
UL Total Traffic [MB/h]
Radius [Km]
1.2
1.4
1
0.8
0.6
0.4
Linear (Light)
150
100
50
0.2
0
0
1
0
25
50
75
100
125
150
175
200
225
250
275
300
Number of users
1.4
Ped-Ind (HSUPA)
Ped-Veh (HSUPA)
Radius [Km]
1
0.8
0.6
0.4
0.2
0
0
25
50
75
100
125
150
175
200
9 10 11 12 13 14 15 16 17 18 19 20 21
Number of users
Fig. 5 Global traffic processed by the Node B in UL for Voice Centric profile.
Fig. 3 Difference between the R99 cell radius for Pedestrian and Indoor
environment, and Pedestrian and Vehicular ones.
1.2
325
225
250
275
300
325
Number of users
Fig. 4 Difference between the HSUPA cell radius for Pedestrian and Indoor
environment, and Pedestrian and Vehicular ones.
Voice Centric
1.4
Data Centric
1.35
1.3
1.25
1.2
1.15
1.1
1.05
1
0.95
10
15
20
25
30
35
40
Number of users
Data Centric
40
20
15
.
Throughput [MB/h]
Voice Centric
25
1.45
35
30
B. Results
Voice Centric
35
10
5
0
10
15
20
25
30
35
40
Number of Users
Data Centric
30
25
20
15
10
5
15
20
100
25
30
35
40
30
35
40
Number of users
80
.
180
70
Voice Centric
160
60
VC, Granted
DC, Granted
50
VC, Queued
40
DC, Queued
30
20
10
0
10
15
20
25
30
Number of users
35
40
90
Data Centric
140
120
100
80
60
40
20
0
10
15
20
25
Number of users
6
Although the HTTP Page Response Time increases while
the number of users gets higher, it still has acceptable values
for some one doing HTTP. Due to OPNETs miss accuracy
Fig. 10 does not give much information. The UL Response
Time should be increasing with more users, and it is not.
However one should remind saturation occurs with only 15
users, and that standard deviation has high values.
Nevertheless the UL response time presents high values for a
100 kB file. The FTP UL Response Time also was simulated,
but once again, due to OPNETs miss function low quality
results were achieved.
This basic 3G network is not the best choice to perform
data based services, especially heavy ones, as FTP, which
needs many available resources in order to have enough
quality to attend client expectations, throughput is very low
and response times are too high compared with other
available technologies.
with the increase in the number of users, means that when the
volume of data based service is higher, network performance
decreases.
300
250
200
150
100
50
0
10
15
20
25
Number of users
Fig. 11 Total traffic in the Node B as a function of the number of users, for
Voice Centric profile.
45
40
35
.
HSUPA gain
30
25
20
15
10
5
0
10
15
20
25
Number of users
IV. CONCLUSION
7
Since this model only regards one single carrier, if more
carriers are available, then it would enhance system capacity.
OPNET Modeler is a simulation tool for communication
networks. The UMTS module based on Release 99 from
3GPP has some limitations, which have impact on results.
Due to the scope of this thesis, the main one is the throughput
limitation. In fact, OPNET only allows 12.2 kbps for
Conversational and Streaming classes, while Interactive and
Background is limited to 64 kbps. ). Another important
limitation is the fact that no CS can be simulated, so voice is
represented by VoIP. Nevertheless it is a very useful
simulator. The network behaviour is studied by increasing the
number of users from 10 to 40 (5 users steps).
Considering the total traffic received by the Node B in UL
(mean value in one hour) one realises it only has a slight
increase until it reaches 20 MB/h. Although, the Voice
Centric average throughput is higher than the Date Centric
one, one must take the high standard deviation values into
consideration, which means those throughput values are not
so different from each other. Low total traffic received can be
justified by the mentioned OPNET limitations. Globally, one
can say that the network starts to saturate for a low number of
users, as seen by the huge percentage of queued requests
comparing to the granted ones.
The comparison between HSUPA theoretical model and
OPNET Modeler R99 simulations for the achievable UL
traffic processed by the Node B is also presented. Of course,
the traffic achieved theoretically is much higher than the one
obtained from simulation, but both approaches tend to
increase with the number of users. The main reason is pretty
obvious: HSUPA is an improvement from basic 3G. Another
important factor is both models have limitations which have
impact on final results. While theoretical model limitations
enhance the outputs, OPNETs downgrade the results.
Although 3G presents good results for some applications, it is
not a good solution for someone who needs an Internet
connection. Nevertheless HSUPA and HSDPA [9] can
provide interesting data rates, and they can be a good solution
when no cable Internet is available.
REFERENCES
[1]
Holma,H. and Toskala,A., WCDMA for UMTS, John Wiley & Sons,
Chichester, UK, 2004
[2] Holma,H. and Toskala,A., HSDPA/HSUPA for UMTS, John Wiley &
Sons, Chichester, UK, 2006.
[3] Lopes,J., Performance of UMTS/HSDPA/HSUPA at the cellular level, M.
Sc. Thesis, IST-TUL, Lisbon, Portugal, Dec. 2007.
[4] Salvado,L. , UMTS/HSDPA Comparison with WiMAX/IEEE802.16e in
Mobility Scenarios, M. Sc. Thesis, IST-TUL, Lisbon, Portugal, Dec.
2007.
[5] Damasso,E. and Correia,L.M., Digital Mobile Radio Towards Future
Generation, COST 231 Final Report, 1999 (http://www.lx.it.pt/cost231).
[6] Correia,L.M., Mobile Communication Systems Course Notes, ISTTUL, Lisbon, Portugal, Mar. 2006.
[7] Esteves,H. and Pereira,M., Impact of intra- and inter-cellular
interference on capacity in UMTS-FDD, Graduation Project, IST-TUL,
Lisbon, Portugal, June 2006.
[8] OPNET Modeler 12.0 PL5 Documentation, OPNET Technologies Inc.,
Bethesda, Maryland, USA, 2007 (http://www.opnet.com).
[9] Antunes,M., Impact of HSDPA Implementation on UMTS Interference
and Cell Coverage, M. Sc. Thesis, IST-UTL, Lisbon, Portugal, Dec 2007.
[10] Sebastio,D., Algorithms for Quality of Service in a WiFi Network, M. Sc.
Thesis, IST-TUL, Lisbon, Portugal, Dec. 2007.