THROUGHPUT ENHANCEMENT USING ADAPTIVE DELAY
BARRIER FUNCTION OVER HSDPA SYSTEM IN MIXED TRAFFIC
SCENARIOS
Yong-Seok Kim
LTE Advanced Tech. Lab., Telecommunication R&D center, Telecommunication Network Business, Samsung Electronics, Korea
Keywords:
Adaptive delay barrier function, a mixed traffic, VoIP, HSDPA.
Abstract:
In this paper, we consider a method to enhance the throughput of HSDPA system in the mixed traffic scenario.
Adaptive delay barrier (DB) function and adaptation method are proposed to maximize throughput of best-
effort (BE) traffic with satisfying the delay latency of voice over internet protocol (VoIP) service. Moreover,
in order to guarantee quality-of-service (QoS) for the mixed traffics, a design of modified scheduling scheme
is provided in this paper. Simulation results show that the throughput of BE traffic service with a channel-
adaptive DB function increases up to 30%, compared with the conventional scheme, without degrading the
capacity of VoIP service over HSDPA system.
1 INTRODUCTION
High-speed downlink packet access (HSDPA), which
supports the peak rate of 14.4 Mbps outperforms
the third generation (3G) WCDMA system speci-
fied in 3GPP Release’99, has been designed to in-
crease the downlink packet data throughput by using
the advanced techniques such as hybrid automatic re-
peat request (HARQ), adaptive modulation and cod-
ing (AMC), fast scheduler in Node B (Holma and
Toskala, 2004). Currently, the transmission of the
real-time (RT) traffic using packet data internet proto-
cols (IPs) is arguably the hottest attention in telecom-
munication technology. This is because it has high
visibility in the consumer space. In other word, it im-
plies cost-efficiency for operator by reason that the
operating, maintaining and updating of circuit-switch
(CS) related part would not be needed anymore. The
challenge for new services that combine voice with
data applications such as video and file sharing is also
popular in today’s research issues (Janevski, 2003).
Scheduling at medium access control (MAC) layer is
a core function to support quality-of-service (QoS)
in these mixed traffic scenario. The basic schedul-
ing algorithm, known as the proportional fairness (PF)
scheduler, designed to support data traffics in 3GPP2
wireless system (Jalali et al., 2000). PF scheduler
provides the effectiveness from the viewpoint of the
multiuser diversity and fairness. However, it does not
able to support RT traffic. In (Ramanan et al., 2001),
modified-largest weighted delay first (M-LWDF) was
proposed to support the stability and throughput while
satisfying QoS requirements of RT traffic user. Al-
though M-LWDF algorithm considers QoS traffic, it
is applied for RT traffic and non-RT traffic separately.
That is, it does not consider various QoS parameter
for multiple traffic scenarios. On the other hand, ur-
gency and efficiency based packet scheduling (UEPS)
was introduced in (Ryu et al., 2005), which designed
to support RT and non-RT traffic user simultaneously
by taking the urgency of scheduling and the efficiency
of radio resource usage into account for. UEPS serves
non-RT traffics until RT traffics approach their dead-
lines, then RT traffics are scheduled with higher pri-
ority during their marginal scheduling time interval,
which is defined as the time-utility function. So it can
maximize the throughput of non-RT traffic with satis-
fying the QoS parameter of RT traffic. However, the
requirement of adaptive adjustment for UEPS scheme
according to the various traffic situations such as dif-
ferent traffic load levels and scheduling urgent factor
was left to the further study issues. Hence, in this
paper, we propose a modified QoS scheduler based
on PF scheme to guarantee QoS in both voice and
127
Kim Y. (2007).
THROUGHPUT ENHANCEMENT USING ADAPTIVE DELAY BARRIER FUNCTION OVER HSDPA SYSTEM IN MIXED TRAFFIC SCENARIOS.
In Proceedings of the Second International Conference on Wireless Information Networks and Systems, pages 127-133
DOI: 10.5220/0002145001270133
Copyright
c
 SciTePress

X
)(n
i
δ
TTI
( )
th
( )
i
n n
X
)(n
i
δ
TTI
( )
th
( )
i
n n
Figure 1: The concept of delay barrier function.
BE traffic service. Moreover, a channel-adaptive de-
lay barrier (DB) function is proposed to increase the
throughput of BE traffic service with satisfying the
delay latency of VoIP service. The performance of
UEPS will be compared with that of our proposed
scheduling scheme. This paper is organized as fol-
lows. In Section 2, we present the scheduling method
in mixed traffic environment. The simulation config-
uration and a criterion to measure the performance of
VoIP service is described in section 3. Finally, simu-
lation results and conclusion are presented in Section
4 and Section 5 respectively.
2 DOWNLINK SCHEDULING
SCHEME
2.1 RF Packet Scheduler for BE Traffic
Service
The use of effective scheduling algorithm is necessary
for improving the throughput of system and keeping
the fairness among users, since the HSDPA system
shares resources with multi-users at the same trans-
mission time interval (TTI). Generally, PF scheduling
algorithm, described in (Jalali et al., 2000), is selected
on HSDPA in order to achieve not only the maximum
system throughput but also the fairness among users.
Using the ratio of the current channel rate to the aver-
age allocated rate, the scheduler chooses a user who
has the maximum output of scheduling metric as fol-
low:
P
(i)
PF
(n) = R
i
(n)/r
i
(n) (1)
where R
i
(n) is user i’s estimated supportable bit-
rate in the n-th TTI and r
i
(n) is the filtered user’s av-
erage throughput, which is updated by a low pass fil-
tering parameter.
Initial
delay barrier
TTI
)(
th
i
n
Maximum
delay barrier
VoIP packet error occurs
One TTI
No CRC or delay error
error
DB
∆
No CRC or delay error
error
Initial
delay barrier
TTI
)(
th
i
n
Maximum
delay barrier
VoIP packet error occurs
One TTI
No CRC or delay error
error
DB
∆
No CRC or delay error
error
Figure 2: The outline of a proposed jump algorithm.
2.2 The Modified Scheduler for VoIP
Traffic Service
Design objective of the PF algorithm is to maximize
long-term throughput for only data service in CDMA-
1x-EVDO system (Jalali et al., 2000). So, it can not
support specific QoS parameter like maximum allow-
able delay for VoIP service. Therefore, for VoIP ser-
vice, time-delay factor should be included in schedul-
ing algorithm. Also, in order to measure delay, the
scheduler puts time-stamps in each packet when it
arrives at the priority queue that is operated with 8
queue buffers to support different QoS service. The
priority would be more weighted as the delay in-
crease. The modified scheduling algorithm operates
by using the output metric as follow;
P
(i)
VoIP
(n) = P
(i)
PF
(n) · f
i
(Que
s
i
,R y
i
) (2)
where P
(i)
PF
(n) is the value of scheduling out-
put metric of user i calculated by using (1). The
delay function f
i
(·) can be designed by f
i
(·) =
r
i
(n) · Que
s
β
i
/R y
γ
i
, where Que s
i
is the accumulated
buffer’s volume of i-th user’s VoIP packets that must
be scheduled at n-th TTI, R y
i
is the remaining delay
latency from the current n-th TTI to the delay bound
at the MAC-hs scheduler. For BE service, f
i
(·) = 1 .
The parameters, β and γ are appropriate weight factor
for each one.
2.3 A Proposed Scheduling Scheme in
Mixed Traffic Scenarios
The QoS provisioning at the radio link means that
the multimedia traffic should get predictable service
form in the wireless system. However, when traffics
demanding different QoS are mixed, it is difficult to
achieve the task. The use of separate priority queues
makes it possible to optimize HSDPA scheduling for
each different QoS service. This paper optimizes the
scheduling algorithm for traffics demanding different
QoS by using priority handling. First of all, VoIP
WINSYS 2007 - International Conference on Wireless Information Networks and Systems
128

0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
Pedestrian A (3km/h) Vehicular A (30km/h)
Vehicular A (120km/h)
Number of VoIP user
Initial
delay
barrier
( )
th
( )
i
n n
Number of VoIP user Number of VoIP user
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
Pedestrian A (3km/h) Vehicular A (30km/h)
Vehicular A (120km/h)
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
0
20
40
60
80
100
10
20
30
40
50
60
7080
90
100
110
120
130
Pedestrian A (3km/h) Vehicular A (30km/h)
Vehicular A (120km/h)
Number of VoIP userNumber of VoIP user
Initial
delay
barrier
( )
th
( )
i
n n
Number of VoIP userNumber of VoIP user Number of VoIP userNumber of VoIP user
Figure 3: The value of initial delay barrier.
flows must be handled by higher priority than inter-
active traffic such as BE service. Generally, VoIP re-
garding conversational service has the highest priority
of traffic classes since it is very sensitive to the trans-
mission delay. Node B receives the value of priority
from the radio network controller (RNC). Hence, we
can differentiate between services through the priority
values. Therefore, a modified proportional fair (PF)
scheduling algorithm is adopted as follows:
P
(i)
Mix
(n) = P
α
i
· P
(i)
VoIP
(n) · δ
i
(n) (3)
where P
i
is the level of priority for i-th user which
provides the high value to VoIP users and a DB func-
tion, δ
i
(n), is represented as shown in Fig. 1, where
n
(i)
th
(n) can be controlled according to the channel
condition of i-th user’s radio link. More detail de-
scription is in next subsection. For BE traffic, δ
i
(·) =
1 and α is appropriate weight factor.
2.4 Throughput Enhancement by using
a Channel-adaptive DB Function
The VoIP traffic has always the highest value of out-
put scheduling metric since the parameter P
i
, indi-
cates the level of priority of i-th user’s traffic class,
may be decided the highest value among the types
of service. Hence, as long as there is a VoIP class
packet in the system, we have a low chance of se-
lecting the BE traffic during the scheduling interval,
and resulting in lower sector throughput for BE traf-
fic user. Therefore, relaxing the level of traffic class
priority should be need if VoIP traffic has some extra
in delay latency, thereby giving more improved sys-
tem throughput. In that point of view, the employ-
ment of adaptive DB function, δ
i
(n), can increase the
throughput of BE traffic service in mixed traffic sce-
nario. That is, it gives the selecting opportunity to
BE traffic packet in a proposed scheduler of (3) by
applying the DB into VoIP traffic only if the VoIP
packet has some marginal time of maximum allow-
able delay latency. For example, if the delay time of
serving VoIP traffic is less than the threshold value of
delay barrier, i.e. n
(i)
th
(n), BE traffic could be served
by a proposed scheduler. On the contrary, if the delay
time of serving VoIP traffic is larger than n
(i)
th
(n), VoIP
traffic would be selected to go service by a proposed
scheduler’s selection rule. In this channel-dependent
adaptive scheme, n
(i)
th
(n) is adjusted adaptively in ac-
cordance with the condition of i-th VoIP service user’s
radio link, which is similar to the processing of the
channel-adaptive outer loop power control (OLPC) in
(Holma and Toskala, 2004), (Koo et al., 2003). A pro-
posed jump algorithm, which is illustrated in Fig. 2,
is operated to maximize the throughput of BE traf-
fic with maintaining the required QoS of VoIP user.
That is, if the error detection device does not detect a
cyclic-redundancy-check (CRC) error of VoIP packet
or a VoIP packet dropped error in a TTI, the value of
delay barrier is increased as follow:
n
(i)
th
(n) = n
(i)
th
(n− 1) + ∆
DB
(4)
where ∆
DB
= Step
size ∗ TargetBLER. Step size
and TargetBLER are any predetermined value. In this
paper, TargetBLER is typically 2% for VoIP service
and Step
size is set by 1. Moreover, the maximum de-
lay barrier should be predefined, so it makes a value of
delay barrier to do not increase above the maximum
delay barrier. In like manner, if there is a CRC error
or a VoIP packet dropped error due to a result of the
exceeding required delay latency in scheuler, n
(i)
th
(n)
is updated to the initial value of delay barrier, which
can be predetermined according to the traffic load of
VoIP users in a sector. It can be determined without
considering of the load of BE traffic since the value of
delay barrier has only to do with the load of VoIP traf-
fic. Fig. 3 illustrates the initial value of delay barrier
to make the percentage of all VoIP users satisfying
the outage condition above 95%. In where, the out-
THROUGHPUT ENHANCEMENT USING ADAPTIVE DELAY BARRIER FUNCTION OVER HSDPA SYSTEM IN
MIXED TRAFFIC SCENARIOS
129

age is defined that VoIP user’s packet error rate (PER)
has been above 2% due to packet loss and packet de-
lay exceeding the target budget. However, when this
channel-adaptive DB function is applied, it makes the
scheduler more flexible in choosing a BE traffic in
mixed traffic scenarios.
3 VOIP SERVICE OVER HSDPA
SYSTEM
3.1 Characteristic of VoIP Service
1) Traffic model and protocol: In this work, two traf-
fic models are considered for the cases of BE service
and VoIP service. In the context of BE traffic, we ap-
plied the full queue traffic model assuming that data
can be always sent when a queue of certain user is
chosen. On the other hand, conversation traffic can be
approximated to the two state Markov traffic model
with a suitable voice activity factor (VAF) (3GPP TR
25.896). The adaptive multirate (AMR) voice codec
is mandatory for voice service in HSDPA systems.
During bursts of conversation, with the AMR mode
of 12.2kbps, the VoIP application generates 32-bytes
voice payload at 20ms intervals (3GPP TS26.236).
During silent periods, a 7-bytes payload carries a si-
lence descriptor (SID) frame at 160ms intervals. A
typical VoIP protocol stack, which employs the real-
time transport protocol (RTP), is encapsulated to the
user datagram protocol (UDP). This, in turn, is car-
ried by IP. The combined these protocols demand a
40-bytes IPv4 header or a 60-bytes IPv6 header. Ob-
viously, the overhead caused from the header to sup-
port VoIP service seriously degrades the spectral ef-
ficiency. Therefore, efficient and robust header com-
pression (ROHC) technique can be used to reduce the
effect of relatively large headers in the IP/UDP/RTP
layers. This technique can reduce the size of the
IP/UDP/RTP headers as little as 2 or 4 bytes. Max-
imum compression 1 byte can be achieved by impos-
ing limitations (IETF RFC 3059, 2001).
2) Definition of VoIP capacity: The VoIP ca-
pacity is in the sense that there exists the maximum
number of VoIP users that can be supported per sec-
tor without exceeding a given outage threshold. In
packet-switch (PS) network, packets will be dropped
under network traffic loads congestion due to packet
loss and packet delay exceeding the target budget. Al-
though some packet loss occurs, the voice quality is
not affected if the amount of packet loss is less than
outage threshold. To proceed with this work, we as-
sume that the PER is kept within 2% and at least 95%
Table 1: Summary of end-to-end delay component for VoIP.
Delay component Delay assumption
Voice encoder 20ms(12.2Kbps)
NodeB scheduling+HARQ Max. 110ms
NodeB site 30ms
ROHC, RLC+MAC process
Downlink propagation
UE scheduling+HARQ 40ms
UE site 30ms
processing, buffering, etc
Uplink propagation
Backhaul delay 30ms
IP network delay About 42ms
of VoIP users should meet the above criterion (3GPP2
TSG-C.R1002-0).
3) End-to-end delay latency for QoS support: To
ensure end-to-end QoS, the low delay is one of the
most important criteria for maintaining high-quality
VoIP service. But, to attain high VoIP capacity,
the scheduler must have sufficient time to manage
voice packets. Of the assumed 285ms end-to-end de-
lay budget for qualified voice service, about 110ms
is available for scheduling in the downlink (ITU-T
G.114). The delay in IP and backhaul network is in
general bounded to 72ms (3GPP TR 25.853), which
is fixed value allowing us to focus on the delay bud-
get within radio access network as shown in Fig. 4.
The end-to-end delay budget in the case of mobile-to-
mobile conversation can be assumed as table 1. Al-
though VoIP performance depends on both downlink
and uplink performance, we would like to set aside
the consideration of both directions as comprehensive
study for future research.
3.2 System-level Simulation Setup
To investigate the performance evaluation with a
mixed voice and BE traffic, a system-level Monte-
Carlo computer simulation is accomplished in this pa-
per. The simulations are carried out with a regular
hexagonal 19 cellular model, where the distance be-
tween Node B is 1km. Mobile terminals should be
uniformly distributed on the 19-cell layout for each
simulation run and assigned different radio link mod-
els according to the assignment probability specified
in (3GPP TS25.101). Note that a realistic model of
the wave propagation plays an important role for the
significance of the simulation results. Shadowing is
modeled by a log-normal fading of the total received
power and a basic attenuation is determined by the
Hata model (3GPP2 TSG-C.R1002-0). Moreover, we
reserve the resources for control and common chan-
WINSYS 2007 - International Conference on Wireless Information Networks and Systems
130

IP Network
UL delay : 40ms
UL Processing delay : 30ms
DL Processing delay : 30ms
DL delay : Variable
Backhaul delay : 30ms
NodeB-GSN:15ms, GSN-NodeB:15ms
IP Network delay : about 42ms
UE NodeB RNC
SGSN/GGSN
Figure 4: End-to-end network delay components for VoIP service.
(a) (b)
(c)
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
(a) (b)
(c)
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
0.5
0.6
0.7
0.8
0.9
1
10 20 30 40 50 60 70 80 90 100 110
Number of VoIP users
Percentage of VoIP Ues with BLER<2%
UEPS
Proposed w/o DB
Proposed w/ DB
Figure 5: Outage probability versus number of VoIP user. (a) Pedestrian A 3km/h; (b) Vehicular A 30km/h; (c) Vehicular A
120km/h.
nel such as OVSF codes and HSDPA power to obtain
the provided simulation results. As mentioned above,
we applied the RTP/UDP/IP packet header compres-
sion using IETF RFC 3059 where the total size of
all compressed header is with 3 bytes (1byte ROHC
base header + 2bytes UDP checksum). Further, the
fast link adaptation and hybrid ARQ (HARQ) mod-
ules are employed in this work. Finally, it has to be
mentioned that we simulate 100,000 TTI snapshots in
average for investigating the performance of the sys-
tem. The main simulation parameters are summarized
in Tables 2.
4 SIMULATION RESULTS
In this section we evaluate the throughput of BE
traffic and the capacity of VoIP with employing a
channel-adaptive delay barrier function in the vari-
ous fading channel environments (3GPP TS25.101).
Also, a proposed scheme is also compared with the
conventional UEPS method. In all simulation, we as-
sume the number of BE traffic users is 40. The per-
centage of users satisfying outage limitation is pre-
sented in Fig. 5 as a function of the number of VoIP
users for different scheduling schemes when the de-
lay latency available to scheduler is fixed as 110ms.
From the figure, we observe that a percentage of VoIP
Table 2: System simulation parameters.
Parameter Assumption
Source traffic AMR 12.2kbps,
VAF=0.32, 2-state Markov,
ROHC 3 bytes [IETF RFC 3059]
Cellular layout Hexagonal grid, 19 sites, 3 sectors
(NB-to-NB 1km)
Carrier frequency 1.9GHz
Propagation loss Path loss=-128.1-37.6*log(R)
Shadowing model Log Normal Std. dev. 8dB,
[Hata model]
UE speed 3km/h,30km/h,120km/h
Antenna gain Node B 14dB/UE 0dB,
Other loss -10dB
Fading Model Pad.A, Veh.A,
Evaluated 3GPP(TS25.101)
UE Rx diversity 2 Rx antennas(UE catagory 10)
Retransmission No RLC retransmission,
Async. HARQ (max. retrial = 6)
CQI delay / error 5TTI (10ms) / 1%
Scheduling Proposed scheme
Reserved 1.Common channel power:20%
2.Associated DCH power per UE:
0.3% of total
3.HS-SCCH power: 9dB offset
over associated DPCH
4.common channel code:10
5.DPCH code per UE:1
6.HS-SCCH codes are considered
THROUGHPUT ENHANCEMENT USING ADAPTIVE DELAY BARRIER FUNCTION OVER HSDPA SYSTEM IN
MIXED TRAFFIC SCENARIOS
131

(a) (b)
(c)
0
1
2
3
4
5
6
0 10 20 30 40 50 60 70 80 90 100
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
0
0.5
1
1.5
2
0 10 20 30 40 50 60 70 80
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
0
0.5
1
1.5
2
0 10 20 30 40 50 60 70 80
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
(a) (b)
(c)
0
1
2
3
4
5
6
0 10 20 30 40 50 60 70 80 90 100
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
0
0.5
1
1.5
2
0 10 20 30 40 50 60 70 80
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
0
0.5
1
1.5
2
0 10 20 30 40 50 60 70 80
Number of VoIP users
Throughput for BE traffic users [Mbps]
UEPS
Proposed w/o DB
Proposed w/ DB
Figure 6: Throughput of BE traffic versus the number of VoIP user for different scheme. (a) Pedestrian A 3km/h; (b) Vehicular
A 30km/h; (c) Vehicular A 120km/h.
Table 3: Summary of the system-level simulation results in
mixed traffic scenario.
Ped.A Veh.A Veh.A
(3) (30) (120)
Proposed 100 90 90
Capacity w/o DB
for VoIP UEPS 90 80 80
traffic Proposed 100 90 90
w/ DB
Proposed 0.63 0.27 0.24
Throughput w/o DB
for BE UEPS 0.95 0.32 0.31
traffic Proposed 1.27 0.37 0.38
(Mbps) w/ DB
users with satisfying PER over outage threshold, i.e.
2%, decrease according to the increase of the num-
ber of VoIP users. This is because the probability
of packet loss due to the exceeding target delay la-
tency becomes more increase with a large number
of VoIP users. However, according to the result, the
more VoIP capacity can be achieved by employing a
proposed scheduling scheme if compared with UEPS
method. For example, when aiming for an identi-
cal percentage of 0.95 in pedestrian radio link en-
vironment, while the UEPS scheme can support 80
VoIP service users, a proposed scheduling scheme
can support 90 VoIP users. In Fig. 6, we evaluate
the achievable BE traffic throughput as a function of
the number of VoIP users when the fixed BE traffic
users of 40. As you can see the figure, the increase
of the number of VoIP users exhibits the decrease of
BE traffic’s throughput remarkably. This is because
the VoIP traffic has the most scheduling opportunity
against BE traffic when the load of VoIP traffic in-
creases. However, note that the throughput with a pro-
posed scheduling scheme can be more achieved if it
is compared with a proposed scheduler without δ
i
(n)
or UEPS. For instance, at the aim of 90 VoIP service
user, the employing channel-adaptive DB enhances
the BE traffic’s throughput as 100% over against a
proposed scheme without δ
i
(n) in (3) and 30% against
UEPS in pedestrian channel condition. We can also
see from the figure, a proposed scheduling scheme
with adaptive DB function still offers better through-
put in vehicular radio link. Table 3 summarizes the
capacity of VoIP and throughput performance in var-
ious propagation conditions. The results confirm that
a proposed scheduler employing a channel-adaptive
DB provides significant improved throughput without
degrading the capacity of VoIP service if compared to
the conventional UEPS.
5 CONCLUSION
In this paper, we propose a modified scheduler em-
ploying a channel-adaptive DB function to enhance
the BE traffic’s throughput in mixed traffic scenar-
ios without degrading the QoS requirement of VoIP
traffic over HSDPA. Our simulation results show that
the capacity of VoIP traffic does not reduced by em-
ploying a proposed scheme as opposed to UEPS. In
addition, the throughput of BE traffic service can be
improved significantly compared to the conventional
scheme. The consideration of the combination of
other traffic types such as web and streaming in the
system may be an interesting issue for future study.
REFERENCES
H. Holma and A. Toskala, 2004. WCDMA for UMTS, John
Wiley and Sons, Third Edition.
WINSYS 2007 - International Conference on Wireless Information Networks and Systems
132

T. Janevski, 2003. Traffic Analysis and Design of Wireless
IP Networks, Norwood, MA, Artech House.
A. Jalali, R. Padovani and R. Pankaj, 2000. Data through-
put of CDMA-HDR a high efficiency-high data rate
personal communication wireless system, IEEE Proc.
VTC2000, 1854-1858.
K. Ramanan, et al., 2001. Providing quality of service over
a sharded wireless link, IEEE Commun. Magazine, 39,
150-154.
S. Ryu, B. Ryu, H. Seo and M. Shin, 2005. Urgency
and efficiency based packet scheduling algorithm
for OFDMA wireless system, IEEE Proc. ICC2005,
2779-2785.
C. -S Koo, et al., 2003. Outer Loop Power Control Using
Channel-Adaptive Processing for 3G WCDMA, IEEE
Proc. VTC2003-Spring, 490-494.
3GPP TR 25.896, Feasibility Study for Enhanced Uplink for
UTRA FDD.
3GPP TS26.236, Packet switched conversational multime-
dia application; Transport protocols.
IETF RFC 3059, 2001. Attribute List Extension for the Ser-
vice Location Protocol.
3GPP2 TSG-C.R1002-0, 2004. CDMA2000 Evaluation
Methodology, version 1.0.
ITU-T G.114, Transmission systems and media - General
characteristics of international telephone connections
and international telephone circuits.
3GPP TR 25.853, Delay Budget within the Access Stratum.
3GPP TS25.101, User Equipment radio transmission and
reception (FDD).
THROUGHPUT ENHANCEMENT USING ADAPTIVE DELAY BARRIER FUNCTION OVER HSDPA SYSTEM IN
MIXED TRAFFIC SCENARIOS
133