Performance Enhancement due to the TNL

Congestion Control on the Simultaneous

Deployment of both HSDPA and HSUPA


Yasir Zaki
1
, Thushara Weerawardane
1
, Andreas Timm-Giel
1
, Carmelita Grg
1
and Gennaro Ciro Malafronte
2

1
University of Bremen, TZI ComNets, Otto-Hahn-Allee NW1, 28359 Bremen, Germany
{yzaki, tlw, atg, cg}@comnets.uni-bremen.de
2
Nokia Siemens Networks S.P.A., Via Monfalcone 1, 20092 Cinisello Balsamo, Italy
{gennaro.malafronte}@nsn.com



Abstractthe main focus of the work presented in this
paper is to analyze the effect of the Transport Network
Layer (TNL) congestion control on the High Speed Packet
Access (HSPA) performance. The TNL and in particular the
Iub link needs to be carefully dimensioned. Firstly because
it has significant impact on the end-to-end and network
performance and secondly due to the high number of
required links in the network, the Iub is a major cost factor
for the network operators. The congestion control function
works together with the air interface scheduler and Hybrid
Automatic Repeat Request (HARQ) in order to control the
offered load to the TNL network. In this manner, the data
flow over the TNL is adequately adapted to the users air
interface data rate and to the available TNL capacity
avoiding congestion in the transport network. In addition,
the paper focuses as well on the effects of the simultaneous
deployment of both High Speed Downlink and Uplink
Packet Access (HSDPA & HSUPA). This is done by
comparing the results from deploying HSDPA or HSUPA
separately in the system against the simultaneous
deployment of both (To the best of our knowledge, there are
only few publications in which this has been investigated,
especially in combination with the Congestion Control). The
reason for such a comparison is to highlight the effects that
appear when both are deployed together, since most of the
previous studies were focusing only on either HSDPA or
HSUPA, whereas the final goal is to use both together in one
system. The simulation results presented in this paper
confirm that the congestion in the transport network can be
controlled in such a way that the available TNL capacity
can be effectively utilized and hence the performance of
HSPA network can be significantly improved in all aspects.
In the ComNets TZI working group at the university of
Bremen, a number of projects focusing on the TNL
dimensioning and TNL features development for the HSPA
network are being worked on [14, 15, 16, 17, 18, 19 and 20].

Index Terms HSDPA, HSUPA, HSPA, Congestion
Detection, Congestion Control

I. INTRODUCTION
HSPA is an evolution of the existing 3GPP WCDMA
Release 99 (R99) standard. It aims to enhance the data
packet traffic transmission on both downlink and uplink.
It significantly improves the system throughput, increases
the system capacity, reduces the delay of the system
compared to R99 system and also can support higher per
user data rates (theoretically up to 14.4 Mbps in the
downlink and 5.76 Mbps in the uplink).
Both HSDPA and HSUPA introduce similar new
features: HARQ is used to recover erroneous air interface
transmissions; a faster scheduling is located at the Node
B which is much closer to the air interface. Moving the
scheduler from the RNC (in R99) allows a better
adaptation to the predicted radio channel and thus a
shorter Transmission Time Interval (TTI) of 2ms is used
for HSDPA and optionally for HSUPA. Additionally
there are still some unique features that each extension
uses separately, for example the Soft Handover (SHO) is
only supported in HSUPA since the other Node B can
receive the UE transmission if in vicinity. In HSDPA the
Adaptive Modulation and Coding (AMC) is used to adapt
the modulation rank and coding rate to the current radio
channel condition of each user, which is not supported in
HSUPA due to uplink power constraints. The HSPA
UTRAN (UMTS Terrestrial Radio Access Network) has
the same network structure as the R99 UTRAN (see
Figure 1).

Figure 1. UMTS R99 network

In order for UMTS R99 to be able to support the HSPA
enhancements, several modifications are required in the
WCDMA R99 physical (PHY) and medium access
JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010 773
2010 ACADEMY PUBLISHER
doi:10.4304/jnw.5.7.773-781
control (MAC) layer. Figure 2 shows the HSPA (HSDPA
R5 [1] and HSUPA R6 [2]) protocol architecture.
PHY
MAC-es
MAC-e
MAC-d
RLC
P
H
Y
M
A
C
R
L
C
PHY
MAC-e
P
H
Y
M
A
C
TNL
EDCH
FP
L
1
-h
s
L
2
H
S
-
D
S
C
H
F
P
L
1
TNL
L
1
L
2
H
S
-
D
S
C
H
F
P
L
2
M
A
C
-c
/s
h
H
S
-
D
S
C
H
F
P
TNL
EDCH
FP
MAC-es
MAC-d
RLC
L
1
L
2
D
S
C
H
H
S
-
F
P
M
A
C
-d
R
L
C
TNL
UE Node B DRNC SRNC
U
u I
u
b
I
u
r
HSUPA
HSDPA

Figure 2. HSPA protocol architecture
II. HSPA TNL CONGESTION CONTROL OVERVIEW
The basic approach of the HSPA Iub congestion
control is to control the Frame Protocol (FP) data rate in
case of congestion. In fact this algorithm provides a
possibility to the upper layers such as RLC and TCP to
recover from congestion by minimizing the number of
transport network packet losses. Further, the algorithm
provides also a mechanism to adaptively reduce the
offered traffic to the transport network according to the
available transport bandwidth under congestion
situations. The rate reduction does not mean degrading
the per-user throughputs but just adapting the MAC data
flows over the Iub to the available TNL bandwidth. All
these activities guarantee the achievable throughput by
reducing the packet losses and thus minimizing the higher
layers retransmissions. Further the congestion control
approach assures that the MAC data flows with the same
statistical congestion detection probability are treated in a
fair way. Fair means that the MAC-d of UEs with
similar radio conditions and throughput has also similar
rate reduction in case of Iub congestion.
The HSPA Iub congestion control consists of two basic
functionalities: congestion detection and congestion
control. The congestion detection is realized in the Node
B for HSDPA and in RNC for HSUPA. This algorithm
can detect any loss of frames or corrupted frames and
such event is considered as the server upcoming
congestion situation. Furthermore, this algorithm also
considers the high delay variation of data flow as
foreseen congestion situation and reacts to avoid such
situation happening in future. In all these situations, the
congestion detection module sends congestion indication
messages to the congestion control module in the Node B
that takes appropriate actions to control the incoming
traffic.
III. CONGESTION DETECTION
The congestion detection can be either reactive or
preventive as discussed in [3]. The preventive congestion
detection scheme tries to identify a possible congestion
before it occurs, whereas the reactive congestion
detection scheme detects the congestion only at its
beginning. In real HSPA system deployments it is
possible to have both reactive and preventive Iub
congestion control schemes working in parallel. This
paper investigates the performance of a fully fledged
HSPA Iub congestion control algorithm in which both
approaches are combined.
The reactive Iub congestion detection mechanism
detects the congestion upon detection of packet losses at
the transport network. The loss can be identified either
due to corrupted frames or due to loss of one or several
complete frames. The later is mostly common due to the
bursty nature of losses. The corrupted frames can be
identified during the payload CRC check. The most
common frame errors in TNL network can be categorized
as follows:
A. Missing a last segment or tail of the FP frame
When the last segment or the tail of the FP PDU is lost,
the receiver waits until it receives the last segment of the
next frame to be reassembled. This results in creating a
large frame with an invalid CRC during the reassembly
process. Both original frames are lost.
B. Missing any segment except the tail of a frame
The receiver can reassemble a FP frame having an
invalid CRC due to a missing segment or due to an
insertion of a foreign segment. Since the tail of the frame
is preserved, the next frame can be detected again.
C. Loss of complete frames
When a burst of cell losses occurs at the transport
network, one FP frame or several FP frames can be lost.
These bursty losses are very common in HSPA networks,
when congestion persists. These losses can be detected by
monitoring the Frame Sequence Number (FSN).
Error type A and B can be easily identified by the FP
payload CRC checksum. Error Type C (Loss of complete
frames) can be identified by using a frame sequence
number (FSN) at the frame header. In both situations in
which the losses can be experienced, the congestion is
considered to be severe and an indication is sent to the
congestion control (CC) module to control the input
traffic. Up to now, the reactive congestion detection is
considered. Next the preventive approach is discussed in
detail.
The preventive congestion detection is in charge of
detecting potential Iub packet losses before they occur.
The preventive based delay build-up algorithm monitors
the FP PDU delay variation for the correctly received FP
frames through the transport network for each MAC-d
flow. The figure below shows the delay variation for
several FP PDU transmissions with respect to the RNC
and Node B reference counters.

Figure 3. FP PDU delay build-up behaviors through TNL network

774 JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010
2010 ACADEMY PUBLISHER
According to Figure 3, the delay build-up algorithm
monitors the TNL delay variations for each arrival of a
FP frame at the Node B. For example, the delay variation
between frames, i and i-1, is given in the following
formula.
1 i 1), - i (i, TNL ) t (
DELAY i
> = A (1)
Where ) t (
i
A is the delay variation between i and i-1.
The accumulated delay variation is given by R(ti):
1 i 1), i (i, TNL ) t ( ) R(t
j
2 i
DELAY i
j
2 i
i
> = A =

= =
(2)
Once the accumulated delay variation exceeds a certain
threshold within a certain time window, congestion is
assumed and an indication is sent to the CC module.
Since it is based on a probabilistic approach, the
indication is considered to be a non severe congestion.
IV. CONGESTION CONTROL
The Iub congestion control is done in the Node B. The
main difference between the HSUPA CC and HSDPA
CC [3] is that the E-DCH congestion control cannot rely
on any rate control provided by the flow control, since E-
DCH has no flow control by its standard. The Congestion
control algorithm (E-DCH_CC) has to monitor the uplink
(UL) E-DCH Iub rate (per MAC-d Flow) in order to
apply Iub UL rate deduction properly. Moreover the E-
DCH_CC supervises the Serving-RLS status (i.e. if the
Node B is the serving or non-serving RLS for the
concerned UE) and it accordingly triggers the specific
congestion control action. The Node B can change from
serving to non-serving and vice-versa without changing
the FP instance and the associated transport bearer.
The congestion control mainly has two states:
congestion control state and the non-congested state
which is called FC state in HSDPA congestion control
and shown in Figure 4.

Figure 4. Flow state machine for CC

At the congestion control state, the CC mechanism
uses the AIMD (a, b) algorithm [4, 5, 6, 7, and 8]. The
CC monitors the available capacity of the TNL network
and controls the incoming traffic flows. In HSUPA, it
decides the allowed capacity over the radio interface for
each user flows that is signaled back to the E-DCH
scheduler in order to decide the uplink grants of the next
TTI for the S-RLS users. In HSDPA, the Node B sends a
CA message to RNC in order to control the offered traffic
to the TNL network. The detailed description of the CC
algorithm can be found in [8].
V. SIMULATION MODEL
The HSPA simulation model is developed using the
OPNET simulation environment [9]. The main purpose of
this model is to perform TNL feature analysis. In addition
the model is also designed to support individual
performance analysis on the protocols like RLC,
TCP/UDP and application layers on a per user level. The
simulation model consists of both HSDPA and HSUPA.
Figure 5 shows an overview of the simulation model. The
red part is the implemented HSUPA protocols whereas
the yellow one represents the HSDPA protocols.
Internet or External
User
Application
TCP
IP
RLC
MAC-d
MAC-es
MAC-e
MAC-d
MAC-e MAC-hs
EDCH
FP
HS-
DSCH
FP
AAL2/ATM

MAC-d
MAC-es
MAC-d
EDCH
FP
HS-
DSCH
FP
AAL2/ATM
RLC
Application
TCP
IP
R
A
B
c
o
n
n
e
c
tio
n
s
S
e
r
v
i
n
g

R
L
S

Node B
UE
RNC

Figure 5. HSPA simulation model overview

In addition, the HSDPA/E-DCH scheduler, SHO,
HARQ, NSRLS (Non Serving Radio Link Set) traffic,
HSDPA/HSUPA congestion control (CC), and HSDPA
flow control were implemented in the simulation model.
In the network configuration, there are three different
traffic types carried in the TNL network: UMTS R99,
HSDPA and HSUPA. In order to handle their different
QoS requirements a traffic separation solution is applied.
A 3 VPs (Virtual Paths) traffic separation solution is
shown in Figure 6.
Node B
RNC

R99 Data Streams
(symmetric)
HSDPA Data Streams
(asymmetric)
HSUPA Data Streams
(asymmetric)
HSUPA User Data (E-DCH)
HSUPA Inband Signalling
HSDPA Inband Signalling
HSDPA User Data (HS - DSCH)
VP
CBR
VP
UBR/UBR+
VP
UBR/UBR+
high priority traffic
ATM Service Category UBR+
ATM Service Category CBR
ATM Network

Figure 6. The three VP traffic separation solution

The three VPs solution is implemented in the
simulation model where each separate VP is assigned to
transport a traffic type but with individual QoS setting. A
CBR (Constant Bit Rate) VP is used for R99; whereas
UBR+ (Unspecified Bit Rate +) VPs are used for HSDPA
and HSUPA since their traffic is treated as best effort
traffic. As R99 traffic mainly consists of delay sensitive
real time traffic like voice, it is configured as a higher
priority over HSPA traffic in order to protect its strict
delay QoS. The impact of different ATM configurations
for traffic separation is analyzed in [12].
VI. TRAFFIC MODELS & SIMULATION SCENARIOS
The packet traffic model is configured at the
application layer of the end user entities. The TCP
protocol is configured with New Reno version. The
traffic model applied in this investigation generates heavy
traffic to be able to monitor the effects best. The traffic
model uses FTP with parameters as shown in table I.
JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010 775
2010 ACADEMY PUBLISHER
Table 1. FTP TRAFFIC MODEL FOR HSPA
FTP Traffic Model (Worst-FTP) Parameters
File size Constant Distribution
Mean file size = 3 Mbyte
Inter-arrival time ~ 0.0 seconds
which means immediately after the first file
downloading of the second file is started

In this FTP traffic model, all users are uploading very
large files continuously all the time. Users have always
data to be transmitted at any time and demanding the
network resources continuously. Therefore such a traffic
model is considered as a worst case traffic from the
network point of view.
The simulations scenarios are classified into three
categories depending on the deployment of the HSPA
traffic (either HSDPA alone, HSUPA alone, or both
combined together). Each category consists of two
scenarios: one is configured with TNL congestion control
called with CC and the other is configured without
TNL congestion control feature denoted as without CC.
All simulation scenarios are shown in table II. The next
section presents the simulation results and analysis.

Simulation scenarios
Config. I Combined HSDPA/HSUPA With CC
Without CC
Config. II HSDPA alone With CC
Without CC
Config. III HSUPA alone With CC
Without CC
VII. CONFIGURATION (I) RESULTS & ANALYSIS
All simulation scenarios run for 1000 seconds. The TNL
network is configured with a last mile rate of 4Mbps.
This in turn is divided into 3 different VPs: Two 2Mbps
UBR+ VPs, one is allocated for the uplink HSUPA traffic
and one for the downlink HSDPA traffic; and one 2Mbps
CBR VP is allocated for R99 traffic.
From the 2Mbps HSDPA and HSUPA VP, 100 kbps
capacity is allocated for the HSUPA and HSDPA in-band
signaling respectively. As for the R99 VP the 2 Mbps are
completely utilized by the R99 traffic for this analysis. In
this context, R99 is considered to be in a worst case
scenario utilizing the full allocated capacity. Following
the main parameter configurations for the simulations are
given:
- TTI = 2 ms HSDPA, 10 ms HSUPA
- Noise Rise = 6dB
- Others-to-own interference factor = 0.6
- Number OF HARQ processes per user flow: 4
- RLC protocol: Operate in RLC AM mode
- TCP protocol: TCP New Reno version
The simulation is performed by configuring six users
in the cell (6 HSDPA in downlink and 6 HSUPA in
uplink). Two scenarios are investigated in this
configuration one with CC and one without. Apart from
the use of the congestion control functionality both
scenarios use the same system and protocol parameter
configurations.
A. ATM link throughput
Figure 7 and Figure 8 show the downlink ATM link
throughput and its probability distribution (CDF)
respectively for the two simulation scenarios. A higher
burstiness of ATM link throughput can be noticed for the
without CC scenario as compared to the with CC
simulation scenario.


Figure 7. HSDPA ATM link instantaneous throughput with/without CC


Figure 8. HSDPA ATM link throughput CDF curve with/without CC

Figure 9 and Figure 10 show the uplink ATM link
throughput and its probability distribution (CDF)
respectively for the two simulation scenarios. Similarly to
the previous results on the downlink, a much higher
burstiness of ATM link throughput is shown for the
without CC scenario as compared to the with CC
scenario.


Figure 9. HSUPA ATM link instantaneous throughput with/without CC
776 JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010
2010 ACADEMY PUBLISHER

Figure 10. HSUPA ATM link throughput CDF curve with/without CC
B. HSUPA Uu Air interface throughput and Noise Rise
The cumulative probability distribution (CDF) of the
overall air interface throughput and the noise rise are
shown in Figure 11 and Figure 12 respectively.
The cell throughput is measured at the Uu interface,
more specifically by taking the sum of the MACe
throughput of all HSUPA users. The system noise rise is
calculated according to:

=
+
=
N
j
j
NR
1
) 1 ( 1
1
n o
(3)
Where is the other to own interference ratio which is
assumed to be a constant value of 0.6, N is the total
number of users in the cell and
j
is user j uplink load
factor which is calculated according to:
W
R
N
Eb
j

+
=
0
1
1
1
n
(5)
Where Eb/N0 is the energy per bit to noise power
spectral density ratio, R is the UE data rate and W is the
chip rate of the HSUPA which is equal to 3.84 Mcps.

The figures depict a large variation of the cell
throughput and the system noise rise for the without
CC scenario compared to the with CC scenario. A
large average cell throughput is achieved by using the
congestion control compared to the other scenario.


Figure 11. HSUPA cell throughput CDF curve with/without CC

Figure 12. HSUPA Noise Rise NR CDF curve with/without CC
C. ATM Cell Discards
The total number of ATM cell discards for the
HSDPA/HSUPA traffic is shown in Figure 13 andFigure
14 respectively. A huge number of ATM cell discards can
be observed for the without CC case. Such effect is
expected since the congestion control algorithm is not
used and the system is congested. This will lead to even
more congestion at the TNL network causing many cell
losses because many retransmissions are triggered by
RLC and TCP layers causing more and more traffic to the
TNL network. Most of these lost packets are recovered
by retransmissions either at the RLC or TCP level; those
retransmissions cause additional traffic to the network.


Figure 13. HSDPA ATM Cells Discard


Figure 14. HSUPA ATM Cells Discard

A clear advantage of using the congestion control
algorithm can be seen from the above figures. The packet
losses at the TNL are significantly reduced which in turn
reduce the higher layer retransmissions and the resultant
TCP end-to-end delays.
D. Application Throughput
The per-user application throughputs for all users as
well as the overall application throughput are shown in
Figure 15, Figure 16, Figure 17 and Figure 18 for both
JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010 777
2010 ACADEMY PUBLISHER
HSDPA and HSUPA respectively. The per-user
throughput is calculated by considering the file size of 3
MByte divided by its upload/download time. It can be
noticed that the per-user/overall application throughput
for the CC based simulation achieves a clear gain
compared to the other scenario for both the uplink and
downlink.


Figure 15. HSDPA per user application throughput with/without CC


Figure 16. HSDPA total application throughput with/without CC


Figure 17. HSUPA per user application throughput with/without CC


Figure 18. HSUPA total application throughput with/without CC
VIII. CONFIGURATION (II) RESULTS & ANALYSIS
The simulation is performed by configuring only six
HSDPA FTP users in the cell (No HSUPA is configured
for the Uplink). The simulation parameters for this
configuration are exactly the same configured for the
previous one. Except the congestion control functionality,
both simulation scenarios use the same system and
protocol parameter configurations.
It can be seen from Figure 19 that there is a clear
advantage of using the congestion control mechanism in
terms of having higher per user application throughput
compared to the case without CC.


Figure 19. HSDPA (separate) per user app. throughput with/without CC

Figure 20 and Figure 21 show the per user/total
application throughput of running HSDPA separate
compared against deploying it with HSUPA. From the
results it can be seen that deploying HSUPA affects the
performance of HSDPA especially in the cases where the
CC algorithm is not used. The reason for that is mainly
because the HSDPA RLC and TCP ACK/NACK(s) that
are carried through the uplink (HSUPA) are getting
delayed and discarded due to the HSUPA load.
Furthermore a congested uplink leads, as explained
before, to delaying the acknowledgments which may
cause HSDPA RLC/TCP to do unnecessary
retransmissions.
What can also be noticed is that having the CC
algorithm activated helps to reduce these effects and to
keep the throughput relatively the same as in the separate
scenario (the green and blue bars in the figures).


Figure 20. HSDPA (separate) compared to HSDPA (combined with
HSUPA) per user app. throughput


Figure 21. HSDPA (separate) compared to HSDPA (combined with
HSUPA) total app. throughput
IX. CONFIGURATION (III) RESULTS & ANALYSIS
The simulation is performed by configuring only six
HSUPA FTP users in the cell (No HSDPA is configured
for the downlink). The simulation parameters for this
778 JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010
2010 ACADEMY PUBLISHER
configuration are exactly the same configured for the
previous one. Except the congestion control functionality,
both simulation scenarios use the same system and
protocol parameter configurations.
Figure 22 shows the per user application throughput of
deploying HSUPA separate with and without the CC
algorithm. The result shows a higher application
throughput for the with CC case.


Figure 22. HSUPA (separate) per user app. throughput with/without CC

Figure 23 Figure 24 show per user/total application
throughput of the separate deployment of HSUPA
compared with the simultaneous deployment with
HSDPA. The results show that deploying HSDPA does
affect the HSUPA performance especially in the without
CC scenario. The reason is similar to the one explained
before in configuration 2, because now the HSDPA flow
in congested which in turn leads to delaying and
discarding the HSUPA RLC/TCP acknowledgments.
Which means that having the CC algorithm in
congested scenarios leads to significant gain in the per
user/overall performance, which can be seen in the green
and blue bar in the figures.


Figure 23. HSUPA (separate) compared to HSUPA (combined with
HSDPA) per user app. throughput


Figure 24. HSUPA (separate) compared to HSUPA (combined with
HSDPA) total app. throughput
X. CONCLUSION
This paper shows the different aspects of the TNL
congestion control algorithm in the HSPA simulation
model. The effects of the TNL congestion control
algorithm for the simultaneous deployment of both
HSDPA and HSUPA performance are presented.
The results showed that using the congestion control
algorithm leads to significant performance enhancements,
ranging from higher per user and total application
throughput, reducing the losses over the TNL network
significantly, lower burstiness and better resource
utilizations by reducing the overall number of
retransmissions.
In addition, the results also showed the effect of
deploying both HSDPA and HSUPA together in one
system. The RLC/TCP acknowledgments that are being
carried by the opposite direction are getting delayed and
in some case even discarded, leading to reducing the
performance slightly when the CC algorithm is in use,
whereas in the case the CC algorithm in inactive the
performance of both HSDPA and HSUPA are
significantly reduced due to TCP mainly because the TCP
acknowledgments that are being carried over the opposite
direction suffers from higher delays and even discards in
the high loaded situations (congested link).
REFERENCES
[1] 3GPP TS 25.309 V6.6.0 (2006-03), Technical
Specification, 3rd Generation Partnership Project;
Technical Specification Group Radio Access Network;
FDD Enhanced Uplink; Overall description; Stage 2
(Release 6).
[2] 3GPP TR 25.855 V5.0.0 (2001-09), Technical Report, 3rd
Generation Partnership Project; Technical Specification
Group Radio Access Network; High Speed Downlink
Packet Access; Overall UTRAN Description (Release 5).
[3] M.C. Necker and A. Weber, Impact of Iub Flow Control on
HSDPA System Performance, Proceedings of the 16th
Annual IEEE International Symposium on Personal Indoor
and Mobile Radio Communications, Berlin, Germany
2005.
[4] S. Floyd, M. Handley, J. Padhye, A Comparison of
Equation-Based and AIMD Congestion Control, ACIRI,
May 2000.
[5] Van Jacobson, Lawrence Berkeley Laboratory & Michael
J. Karels, University of California at Berkeley, Congestion
Avoidance and Control, November, 1988.
[6] Sagfors, M. Virkki, V. Kuningas, T., Overload Control
of Best-Effort Traffic in the UTRAN Transport Network,
VTC 2006 conference, May. 2006, Melbourne, Vic.
[7] Sz. Ndas, S. Rcz, Sz. Malomsoky, S. Molnr:
Connection Admission Control in the UTRAN Transport
Network, Telecommunication Systems, 28(1): 929,
2005.
[8] Thushara Weerawardane, Andreas Timm-Giel, Gennaro C.
Malafronte, Durastante Gianluca, Stephan Hauth,
Carmelita Grg., Preventive and Reactive based TNL
Congestion Control Impact on the HSDPA Performance
VTC IEEE conference, May 2008, Singapore.
[9] OPNET technologies Inc., OPNET Modeler Accelerating
Networks R&D, www.opnet.com (Last accessed in
04.11.08)
[10] Harri Holma and Antti Toskala. WCDMA for UMTS:
Radio Access for third Generation Mobile Communication.
Third Edition. John Wiley & Sons 2004.
[11] Harri Holma and Antti Toskala. HSDPA/HSUPA for
UMTS: High Speed Radio Access for Mobile
Communication. John Wiley & Sons 2006.
JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010 779
2010 ACADEMY PUBLISHER
[12] X. Li, R. Schelb, C. Grg and A. Timm-Giel, "UMTS
HSPA and R99 Traffic Separation," in 10th IFIP
International Conference on Mobile and Wireless
Communications Networks (MWCN), Toulouse, France,
Sept. 30-Oct. 2, 2008,
[13] Yasir Zaki, Thushara Weerawardane, Xi Li, Andreas
Timm-Giel, Gennaro Ciro Malafronte and Carmelita Grg:
Effect of the RLC and TNL Congestion Control on the
HSUPA Network Performance, Mosharaka International
Conference on Communications, Computers and
Applications MIC-CCA 2008 in Amman-Jordan
[14] T.L. Weerawardane, A. Timm-Giel, C. Grg, T. Reim:
Performance Analysis of the Iub Interface in UMTS
Networks for HSDPA, Mobilfunk Technologien und
Anwendungen, 10. ITG-Fachtagung, Juni 2005,
Osnabrck.
[15] L. Zhao, T.L. Weerawardane, A. Timm-Giel, C. Grg, U.
Trke, M. Koonert: Overview on UMTS HSDPA and
Enhanced Uplink (HSUPA), Mobilfunk Technologien und
Anwendungen, 11. ITG-Fachtagung, Mai 2006,
Osnabrck.
[16] Thushara Weerawardane, Andreas Timm-Giel, Carmelita
Grg, Thomas Reim, "Impact of the Transport Network
Layer Flow Control for HSDPA Performance", IEE
conference, Sep. 2006, Colombo, Sri Lanka.
[17] X. Li, R. Schelb, C. Grg and A. Timm-Giel,
"Dimensioning of UTRAN Iub Links for Elastic Internet
Traffic." International Teletraffic Congress, Beijing,
Aug/Sept. 2005, 2005.
[18] X. Li, R. Schelb, C. Grg and A. Timm-Giel,
"Dimensioning of UTRAN Iub Links for Elastic Internet
Traffic with Multiple Radio Bearers," in Proc. 13th GI/ITG
Conference Measuring, Modeling and Evaluation of
Computer and Communication Systems, Nrnberg, March
2006, 2006
[19] X.Li, W. Cheng, A.Timm-Giel, and C. Grg, "Modeling
IP-based UTRAN for UMTS in OPNET", distinguished
paper award, in Proc. OPNETWORK 2007, September,
2007, Washington DC, USA, 2007
[20] X.Li, Y.Zeng, B. Kracker, R.Schelb, C.Grg and A. Timm-
Giel, Carrier Ethernet for Transport in UMTS Radio
Access Network: Ethernet Backhaul Evolution, (accepted
for publication) 2008 IEEE 67th Vehicular Technology
Conference VTC2008-Spring, May 2008, Singapore, 2008.




Yasir Zaki received his bachelor (B.Sc.) degree in
Electronics and Communication at the University of Baghdad,
Iraq in 2004 and master degree (M.Sc.) in Communication and
Information Technology at University of Bremen, Germany in
2007.
After completing his bachelor degree he was awarded a
DAAD scholarship to finish his master studies in Germany.
Then after completing his master degree he joined the Center
for Computer Science and Information Technology (TZI) of the
University of Bremen in the Communication Networks group as
a scientist researcher and a PhD candidate in 2007. He worked
in the industrial research project funded by Nokia Siemens
Networks on performance optimization of UMTS/HSPA radio
networks and transport networks from 2007 to 2008. Starting
from 2008 he started working in the 4WARD European project
that is focusing on the Future Internet; where he is mainly
focusing on the network virtualization. Currently he is
investigating how the LTE (Long Term Evolution) system can
be virtualized, the wireless resources in particular and how this
could be shared between multiple virtual operators.
Mr. Yasir Zaki has published a number of scientific papers in
the field of communication networks.



Thushara Lanka Weerawardane received bachelor (B.Sc.)
degree in Electrical Engineering at the University of Moratuwa,
Sri Lanka in 1998 and master degree in Communication and
Information Technology at University of Bremen, Germany in
2004.
He worked as an assistant network manager in the
department of electrical engineering and as a system engineer
for Lanka Educational and Research Network (LEARN) in Sri
Lanka from 1999 to 2002. After completing his master degree,
he joined Center for Computer Science and Information
Technology (TZI) of the University of Bremen in the
Communication Networks group as a scientist researcher in
2004. He led the industrial research project funded by Nokia
Siemens Networks on performance optimization of
UMTS/HSPA radio networks and transport networks from 2004
to 2007. His main responsibilities are UMTS/HSPA based
protocol design and development, HSPA (HSDPA/HSUPA)
simulator design and development for performance analysis of
HSPA radio networks and transport networks, optimization of
the HSPA transport network (Iub/Iur) by deploying different
congestion control and flow control features. Currently, he is
leading an LTE simulation research project funded by Nokia
Siemens Networks. In this ongoing research project, LTE
system level simulator design and development for the
optimization of LTE transport networks are the main
consideration.
Mr. Thushara Weerawardane has published many scientific
papers in the field of communication networks and is a member
of IEEE.



Andreas Timm-Giel, Dipl.-Ing in Electrical
Engineering/Information Technology (EE/IT) at University of
Bremen, Germany, 1994, Dr.-Ing. (PhD) in EE/IT on radio
channel modeling, University of Bremen, 1999.
From 1994 1999 he led a group at the University of
Bremen, that participated in several European R&D projects on
mobile and satellite communications. Starting in January 2000
he joined Media Mobil Communication GmbH as Technical
Project Leader of the EU funded project SATISFY2000. He was
involved in the technical and commercial set up of the mobile
satellite network and service provider M2sat Ltd. as Technical
Product Manager and Manager Network Operations. In
December 2002 he joined the Communication Networks Group
at the University of Bremen as senior researcher and lecturer.
He is leading several industry, national and EC funded research
projects at the university. Since October 2006 he is additionally
directing the interdisciplinary concerted activity Adaptive
Communications of the Center for Computer Science and
Information Technology (TZI) in Bremen. His research interests
are adaptive mobile and wireless communication and sensor
networks.
Dr Timm-Giel is author or coauthor of 10 book contributions
and more than 65 reviewed publications in journals and on
international conferences. Dr. Timm-Giel is frequent reviewer
and TPC member for international conferences and journals and
is Member of IEEE and VDE/ITG.


780 JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010
2010 ACADEMY PUBLISHER

Carmelita Grg received her diploma degree from the
Department of Computer Science, University of Karlsruhe and
the Dr. rer. nat. degree and the appointment as lecturer from the
Department of Electrical Engineering, Aachen University of
Technology.
From 1985 until 1989 she worked as a consultant in the field
of communication networks. Since 1989 she has been working
as a group leader and since 1997 as an Assistant Professor at the
Communi-cation Networks Institute, Aachen University of
Technology. Since 1999 she is leading the Communication
Networks Group (ComNets) at the University of Bremen within
TZI (Center for Computer Science and Information
Technology) and MRC (Mobile Research Center). Her research
interests include: Performance Analysis of Communication
Networks, Stochastic Simulation, Rare Event Simulation, High
Speed Networks, Personal Communication, Wireless Networks,
Mobility Support, New Services and Applications in
Telecommunication Networks, Network Virtualization.
Prof. Grg has been active in European projects starting with
the RACE program. She has been an evaluator and auditor for
the European Commission. The research group in Bremen
consists of about 15 Ph.D. students / research assistants, which
are funded by the state of Bremen and third-party projects
(European projects, DFG projects, BMBF projects, bilateral
industry projects). Prof. Grg has published a large number of
scientific papers in the field of communication networks. She is
a member of the board of the ITG (Information Technology
Society, Germany) and speaker of the ITG working group 5.2.1
on "System Architecture and Traffic Engineering".



Gennaro Ciro Malafronte received master degree in
Electronic Engineering in 1999 and PhD in Electrical
Engineering on Electrical and Electronic Measurements in 2003
at the University of Naples Federico II.
In 2001, he joined the Italian COM division of the Siemens
Corporation, then merged in the Nokia Siemens Networks in
2007. He started his activity at Siemens COM for concept
analysis of Smart Antennas and Software Radio Technology.
He then worked as Sales Support for some UMTS service
providers in Europe. He was involved in special projects in
collaboration with the European Space Agency (ESTEC) for
requirement specifications of satellite transmission paths in the
terrestrial Siemens UMTS systems. Since 2004 he joined the
Siemens System Engineering Transport Expert Group and he
focused his activity in the optimization of the transport features
of UMTS terrestrial interfaces (Iub,Iur, Iu). In that field he was
responsible for the definition of HSDPA/HSUPA Iub/Iur
congestion control algorithms and he contributed to the 3GPP
for the definition of standardized procedures for Iub/Iur
congestion control.
Dr. Gennaro Ciro Malafronte holds a number of patents in
the field of Iub/Iur HSDPA/HSUPA congestion control as
Method for congestion control with macro diversity (EP 1
816 879 A1), Controlling congestion over a non-serving
branch (EP 1 901 493 A1), Method for decoupling congestion
control in a cascade of network elements of an UMTS radio
access network (WO 2008/064983 A2) Controlling
congestion detection in HSDPA systems (WO 2008/037357
A1), He is co-author of a number of papers in the field of
Iub/Iur HSDPA/HSUPA congestion control. His current field of
interest is Transport Optimization for LTE systems.

JOURNAL OF NETWORKS, VOL. 5, NO. 7, JULY 2010 781
2010 ACADEMY PUBLISHER