QPF SCHEDULING SCHEME FOR PERFORMANCE

IMPROVEMENTS OF INTEGRATED MULTIMEDIA

APPLICATIONS OVER 3.5G NETWORK

Shin-Jer Yang, Li-Chih Lin

Dept. of Comp. Science & Infor. Management, Soochow University, No. 56, Sec. 1, Kuei-Yang St., Taipei, Chinese Taiwan

Yung-Ming Hsieh

Department of Accounting, Soochow University, No. 56, Sec. 1, Kuei-Yang St., Taipei, Chinese Taiwan

Keywords: VoIP, HSDPA, HARQ, QPF (QoS-based Proportional Fairness), PF (Proportional Fairness).

Abstract: IMS can provide full IP technology platform which combines information and communications

technologies to achieve the objectives of integrated multimedia service. In IMS, SIP is introduced to

transmit session control signal. Currently, the session transmission time of SIP is affected by wireless

channel bandwidth, Frame Error Rate (FER), message exchange volume, retransmission mechanism and

other factors. In order to effectively reduce the delay time for session transmission, we did revise

conventional MAC-hs scheduling algorithm (i.e. PF) and propose QoS-based Proportional Fairness

algorithm, called QPF, to enhance the session setup on SIP with channel performance of wireless

communication. We perform simulations with comparisons on two KPIs of Delay and Throughput between

PF and QPF schemes, by using NS-2 software tool under various multimedia applications, such as VoIP,

VoD and Web. The simulation results indicate that QPF lowers 20% for VoIP delay and obtains 8.19%

more VoIP Throughput than PF. The QPF performance for VoD delay can be enhanced by 11.53%, the

Throughput is increased by 8.05%. Also, the delay for Web of QPF can be improved by 28.58%, and the

Throughput is enhanced by 25.25%. Consequently, the proposed QPF can enhance transmission

performance and obtain a higher utilization of system resources for various multimedia applications in

wireless network.

1 INTRODUCTION

IP Multimedia Subsystem (IMS) is the system used

to control the multimedia application to be proposed

by Release 5 (R5) of 3GPP. In IMS, the session

transmission time for SIP (Session Initiation

Protocol) refers to the session setup time plus VoIP

data transfer time starting from User Agent Client

(UAC) call to the receipt of ring back tone, in which

the delay time can be divided into two parts: (1)

network transmission time which refers to time spent

for data transmission from one node to another,

including transmission and retransmission time and

(2) the delay time on network handling which is

affected by Radio Link Control (RLC) and error

detection. RLC signals will be affected by Multipath

Fading, noise interference and buildings while

transmitting over the wireless network. Also, this

will lead to worse wireless channel conditions and

high error rate in wireless link layer. In order to

reduce the impact of high error rate of wireless

channels, this paper provides a more reliable data

transmission service for the upper layer, with some

guaranteed data transmission quality mechanisms

such as Radio Link Protocol (RLP) introduced in the

wireless link layer (Fathi et al., 2004).

In order to effectively control the wireless link

layer over 3.5G network, a new sub-layer for MAC-

hs (high speed) is added on the Media Access

Control (MAC) layer of Node-B, in which the

features of Fast Scheduling and Hybrid Automatic

Repeat Request (HARQ) will effectively help reduce

the delay time for network transmission, with

different services quality that are defined by diverse

media application demands as per delay time. The

main purpose of this paper is to study and explore

Yang S., Lin L. and Hsieh Y..

QPF SCHEDULING SCHEME FOR PERFORMANCE IMPROVEMENTS OF INTEGRATED MULTIMEDIA APPLICATIONS OVER 3.5G NETWORK.

DOI: 10.5220/0003428500050012

In Proceedings of the 13th International Conference on Enterprise Information Systems (ICEIS-2011), pages 5-12

ISBN: 978-989-8425-56-0

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

the scheduling enhancement on 3.5G wireless

network transmission to achieve Quality of Service

(QoS) and performance improvement of multimedia

application service.

The remainder of this paper is organized as

follows. In Section 2, related literature review and

discussion proceeded on the key technologies and

applications of HSDPA and on new MAC-hs

protocol. In Section 3, we will consider QoS in

terms of PF in HSDPA and propose QPF (QoS-

based Proportional Fairness) with scheduling to be

designed to improve its retransmission and

scheduling mechanism. In Section 4, we plan and set

up simulation environment and list the execution

procedure. In Section 5, we present and analyze the

simulation results. In Section 6, we make the

conclusion and indicate the future research direction.

2 PRELIMINARY AND RELATED

WORKS

2.1 Integration of IMS and 3G

Network

With the prevalence of communication system and

Internet as well as evolution of new technologies,

the integration of IP network and multimedia service

has become a new trend for future development.

IMS provides a good solution about how to attain

real-time communications and multimedia

applications over 3G networks (Kara & Planat,

2007). Hence, IMS is a technology platform upon

SIP and is irrelevant to access network. Under a

3GPP IMS structure, Call Session Control Function

(CSCF) is a primary component which is responsible

for the control of SIP-based voice and multimedia

conversation. Therefore, CSCF feature is described

as followings and its structure is shown in Figure 1

(3GPP, 2010).

a) Proxy CSCF (P-CSCF) is the entry point of IMS.

Also, it is responsible for subscriber network

searching, and providing security and authentication

functions.

b) Serving CSCF(S-CSCF) takes charge of phone

service and session control, which is the control core

of entire IMS.

c) Interrogating CSCF (I-CSCF) is used for every

calling to select corresponding S-CSCF.

In Figure 1, Serving GPRS Support Node (SGSN)

and Gateway GPRS Support Node (GGSN) of

GPRS pass internal IP network of 3G to provide

packet service for communication signal, in which

the message flow for communication with traditional

telecom network enters PSTN/ISDN via Media

Gateway (MG) under the control of IMS. S-CSCF

controls MG operation through Media Gateway

Control Function (MGCF), while Signal Gateway

(SG) supports Signalling System 7 (SS7) transmit

over IP network and provides signalling service

between IMS and traditional telecom network.

Figure 1: Basic infrastructure for the integration of IMS

and 3G network.

2.2 Session Transmission Time of SIP

The session transmission time of SIP refers to SIP

session setup time plus VoIP data transfer time. The

session setup time was spent on the INVITE request

sending from User Agent Client (UAC) to User

Agent Service (UAS) which is informed of receipt

of this session call from the Server (i.e. UAS

receives the request of Acknowledgement (ACK).

As a consequence, SIP time can be regarded as a

total of completion of all message processing and

delay time for VoIP data transmission between

client-side and server-side. The delay for session

setup involves many factors, in which the

transmission delay over network is the major part

including retransmission time delay and queue delay

caused by FER. The procedure of SIP session

transmission is shown in Figure 2, and the main

steps are described as follows:

a) UAC issues an INVITE request carrying with

media negotiation message at the beginning.

b) When UAS receives the invite request, it

responds the session of 183 Session Progress.

c) When UAC receives the response, it issues

PRACK temporary request as per UAS’s media

capability.

d) UAC proceeds to the verification with 200 OK.

e) UAS starts to issue 180 ringing message when

resource reservation of UAS side is done.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

f) UAS will issue 200 OK to respond to the

primitive INVITE request.

g) UAC further confirms with ACK to be a final

response to the INVITE request and then proceeds

VoIP data transmission.

Figure 2: Delay time for SIP session setup.

2.3 Key Technologies of HSDPA

High Speed Downlink Packet Access (HSDPA),

namely the 3.5G network, is an enhanced wireless

technology of Wideband Code Division Multiple

Access (WCDMA) (Krause, 2002). The objective of

HSDPA technology is to slightly adjust or upgrade

the software without changing current 3G/WCDMA

network structure, to enhance Peak Data Rate of

Downlink transmission packet data to over 10 Mbps

and to shorten the packet latency for allowing the

system to transmit packet data more efficient.

HSDPA technology lists diverse features

including New MAC-hs and Channel, HARQ and

Fast Scheduling to be compared to WCDMA (Ghosh

et al., 2004). These features are illustrated as follows.

(a) New MAC-hs and Channel: HSDPA defines a

new MAC sub-layer as MAC-hs (3GPP, 2001) on

Node-B, where Node-B is a term used in UMTS

equivalent to the BTS (base transceiver station), with

a structure as shown in Figure 3. The MAC-hs takes

charge of directing and distributing any a requested

User Equipment (UE) for sharing channel resource of

HSDPA and provides scheduling task of High Speed

Downlink Shared Channel (HS-DSCH). The

Transmission Time Interval (TTI) of HSDPA is 2 ms.

It obtains five times growth comparing to TTI (10 ms)

of Release’99 specification and which leads to an

achievement of fast scheduling. The four functions of

MAC-hs include: (1) Flow control; (2) Scheduling

with Priority handling; (3) HARQ; and (4) Transport

Format and Resource Combination (TFRC) selection.

Figure 3: Structure diagram of MAC-hs at URTAN side.

(b) HARQ: The HARQ is an error correction

technology that is frequently used in communication.

It is a technology combining both Forward Error

Correction (FEC) and Automatic Repeat reQuest

(ARQ). Also, it can swiftly adjust the transmission

rate of channel to achieve the combinations of FEC

and retransmission according to the link conditions.

HSDPA is directly controlled by MAC-hs of Node-

B instead of by RNC (Ameigeiras et al., 2004). The

scheduling algorithms of HSDPA were normally

divided into three types (Jalali et al., 2000): (1)

Round Robin (RR), (2) Max C/I and (3) Proportional

Fair (PF). When considering the shortest delay time

for real-time voice transmission and network

fairness of QoS, we choose to revise PF scheduling

algorithm. User who gains an improvement on

Channel Quality Indicator (CQI) will be able to

benefit from higher priority rendered by PF. PF

scheduling adopts an updating mechanism on

average speed rate to avoid starvation phenomenon

and to improve system delay time (Aniba & Aissa,

2004). In PF algorithm, each user will be given a

corresponding equation with priority level P

shown below:

(1)

In Equation (1), DRC represents Data Rate

Control while DRC

(t) refers to data rate supported

by scheduled user i (Total N users) at time t

according to current channel carrier to interference

QPF SCHEDULING SCHEME FOR PERFORMANCE IMPROVEMENTS OF INTEGRATED MULTIMEDIA

APPLICATIONS OVER 3.5G NETWORK

ratio. The updated equation of average transmission

rate R

(t) introduced by user i as Equation (2).

( t + 1) = (1 - 1/tc) * R

(t)

+ 1/tc * Current_Rate

(2)

In Equation (2), where tc represents the length of

Time slot while Current_Rate refers to the current

frame transmission rate.

3 RETRANSMITTING

AND SCHEDULING

IMPROVEMENTS IN HSDPA

Retransmitting and scheduling of HARQ in HSDPA

will be added on MAC-hs sub-layer, the major

consideration lies in swiftly and clearly controlling

usage conditions for radio resource, allowable for the

packet to quickly transmit without any delay in order

to enhance transmission efficiency. According to the

conventional 3GPP, QoS parameters of Iub (Interface

between RNC and Node-B) in HSDPA include

(Holma & Toskala, 2006): (1) Scheduling Priority

Indicator (SPI): the value range of SIP is [0, 1,…,

15], and the larger the number, the higher the priority

level; (2) Guaranteed Bit Rate (GBR); and (3)

Discard Time (DT). Also, we will categorize the

media service types into Conversational, Streaming,

Interactive and Background services by SIP priority

upon the sensitivity level of services on the delay and

diverse media service flow applications (Dixit et al.,

2002; Li & Aaron Gulliver, 2009). The fundamental

characteristics and application examples of various

service types are summarized in Table 1.

Table 1: Class of media service in wireless network.

3.1 Design Issues of QPF Scheduling

The QPF is proposed to revise the scheduling and

priority processor of MAC-hs based on HSDPA

scheduling, using SPI indicator to determine those

conversation types (e.g. VoIP) and higher FER

frame which requires retransmission as higher

scheduling weight than other service flows, followed

by streaming type, interactive type, and background

type. The operational flow of QPF scheduling

processing is shown in Figure 4 and its descriptions

is illustrated in the following steps.

Figure 4: Operational flow of QPF scheduling.

Step 1: Calculation and Weight Setting of

Priority Level

In Equation (3), W

represents weight of service type

of that subscriber, with rest of variables subject to

Equation (1). In addition, the multimedia application

type of 3G network is conformed to QoS needs, so

that the weight of service type W

is set as following:

4 for Conversational, 3 for Streaming, 2 for

Interactive and 1 for Background.

(3)

Step 2: Priority Value Queue

Place pri value of UE which stays synchronous with

data transmission accompanied by Dedicated

Physical Channel (DPCH) into the pri queue.

Step 3: Resource Allocation

Gather allocable real resources of Node-B.

a) Determine whether pri queue is empty and then

judge whether there are usable HS-DSCH resources.

The resource allocation process will come to an end

if there are usable non-link channel resources or

vacant pri queue.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

b) Calculate occupied real resources to be necessary

and send Modulation and Coding Scheme (MCS) for

UE according to retransmission request if HARQ

requires a resending.

c) Dispatcher will firstly transmit a retransmission

required for HARQ packet and calculate priority

(pri) value as Equation (3).

d) Choose the user who gains highest pri value to

execute dispatching, and then calculate average

service rate of each separate users as Equation (2).

3.2 QPF Algorithm Design

Based on QPF operations as depicted in Figure 4, we

design the QPF algorithm and list its pseudo codes

as follows.

Algorithm QPF( )

{

Input: #Traffic flow of HSDPA UE

#include "hsdpalink.h" //MAC-hs

implement-ation (also includes

physical layer)

#include "cqi.h" //Lookup table for CQI

to MAC-hs PDU size

#include "error_model.h" //Reads

tracefiles for CQI and transmission

power

#define FLOW_CLASS(): //Traffic flow

classification and add weight

#define FLOW_CHECK(): //Check for

activated flows

#define SORT_POWER(): //Sort by Maximum

relative CQI

#define HARQ: //HSDPA Hybrid Automatic

Repeat Request

#define CACULATE_DRC: //Calculate HSDPA

LINK Data Rate Control

#define struct queue(); //define a

priority queue structure

static vector < float > power_; //HSDPA

UE Power value

boolean retransmit_flag = true; //data

retransmit indicator

char[15] flow_type

int weight

int flow_max

OUTPUT:// Loop through the priorities

of each flow, in sort order

if ( priority is Maximum)

{ Transmission the package }

else

{ waiting in queue buffer }

---------------------------------------

------------------------

Method:

BEGIN

{

Set weight = FLOW_CLASS();

Execute FLOW_CHECK();

for (int i = 0; i < queue.size();i

++) {if ( power_.at(i) > 0.0 )

{ //channel resource enough

if ( retransmit_flag )

{ EXEC HARQ ; break; }

{power_at(i)=

power_.at(i)* weight;}

} else { power.push_back(0.0); }

}

EXEC SORT_POWER();

EXEC CACULATE_DRC;

}

End

---------------------------------------

------------------------

Procedure FLOW_CLASS()

{

if (flow_type = ”conversational”)

{set weight = 4}

else if (flow_type = ”streaming”)

{set weight = 3}

else if (flow_type = ”interactive”)

{set weight = 2}

else if (flow_type = ”background”)

{set weight = 1}

Return weight;

}End FLOW_CLASS()

---------------------------------------

------------------------

Procedure FLOW_CHECK()

{

for (int i = 0; i < flow_max; i ++) {

if ( activated_.at(i) )

{ get power for this flow

insert received power into

queue}}

Return queue()

}End FLOW_CHECK()

---------------------------------------

------------------------

Procedure SORT_POWER()

{

float pow_val;

for ( int i = 1; i < queue_.size();

i ++ ) {

pow_val = power_.at(i);

int j = i - 1;

while ( j >= 0 && power_.at(j) <

pow_val )

{power_.at (j + 1) =

power_.at(j);

j --; }

power_.at(j + 1) = pow_val;}

} End SORT_POWER()

---------------------------------------

------------------------

}

END QPF.

QPF SCHEDULING SCHEME FOR PERFORMANCE IMPROVEMENTS OF INTEGRATED MULTIMEDIA

APPLICATIONS OVER 3.5G NETWORK

4 ENVIRONMENT SETUP

AND PROCEDURE OF

SIMULATIONS

For the purpose of simulation experiments, this paper

utilizes network simulator of NS-2.28 software

version with insertion of EURANE-1.11 (Enhanced

UMTS Radio Access Network Extensions) module

to perform network simulations.

4.1 Setup of Simulation Environment

HSDPA network takes Dedicated Channel (DCH)

and HS-DSCH into consideration. They are

employed to signal control and data transmission,

respectively. As a result, the status of packet

transmission of these channels can be monitored to

improve transmission efficiency of network. The

network topology of simulation environment is

shown in Figure 5.

Figure 5: Topology of simulation environment.

Table 2: Settings of Simulated Parameters.

Service Types Application Traffic Source

Peak

Sending Rate

Streaming VoD CBR 128 kbps

Conversational VoIP Expo. On/OFF 30 kbps

Interactive Web Pareto On/OFF 60 kbps

Background FTP FTP 3434 kbps

In the experiment, this paper chooses a previous

network application (Shreevastav et al.,2009) to

stand for each media service flow: (1)

conversational type: VoIP uses Exponential On/Off

to generate data flow. (2) streaming media type:

VoD uses CBR to simulate. (3) interactive type:

Web uses Pareto On/Off to generate data flow. (4)

background type: FTP application uses embedded

FTP traffic generator of NS-2 to create information

flow. The above related parameters setting is shown

in Table 2. Because Nodes of HSDPA must be

precisely configured before simulations, the

configuration parameters of these nodes including

bandwidth and delay are shown in Table 3.

Table 3: Parameters Setting of Nodes.

Starting node Terminal node Bandwidth Delay

Serv. node B Serv. node A 10 Mbps 35 ms

Serv. node A GGSN 10 Mbps 15 ms

GGSN SGSN 622 Mbps 10 ms

SGSN RNC 622 Mbps 0.4 ms

RNC Node-B 622 Mbps 15 ms

4.2 Procedure of Simulation

The core network topology is composed of GGSN,

SGSN and RNC as shown in Figure 5. It will be

allocated with different application service flows (2

VoIP, 2 VoD, 2 Web) by UE side. The FTP can be

regarded as background flow. It will not only

increase the number of FTP UE to measure the

results of Delay and Throughput which pass the core

network, but also compare their differences and

analyze experimental results between PF and QPF

schemes.

5 SIMULATIONS RESULTS

AND ANALYSIS

In this section, we take Transmission delay and

Throughput as KPIs (Key Performance Indicators)

to measure the performance.

1. Transmission delay refers to time spent on

network transmission from UE side to the server. To

measure the results according to the transmission

delay as defined by RFC 3550, two parameters will

be derived when a voice packet (i) is received by the

receiving side. They are packet timestamp Ri and

packet RTP timestamp Si. Therefore, the

transmission delay D(i, j) between any two packet i

and packet j can be determined by Equation (4).

D(i, j) = (Rj - Ri) - (Sj - Si) (4)

The experimental results as shown in Figure 6

indicate that QPF generally gains better efficiency

than PF under light or heavy network traffic when

there is a sufficient bandwidth. Practically, the QPF

holds stable delays upon various application service

flows. Therefore, simulation results demonstrate that

the delay time can significantly decrease by taking

the advantages of QPF algorithm.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems

Figure 6: Simulation results in Delay.

2. Throughput is the average rate of successful

packet delivery between two nodes of network. It is

usually measured in bits per second (bit/s or bps),

which is calculated as Equation (5).

Throughput = received bits / T (5)

In Equation (5), received bits represent the

amount of bits t5o be received at all destination

nodes during a certain period of time T.

According to the experimental results as depicted

in Figure 7, the Throughputs of QPF are very close

to those of PF under light network traffic (UE =<

11), but the QPF gains higher Throughput than PF

under heavy network traffic (UE > 11). Therefore,

simulation results indicate the Throughputs of QPF

are better than PF in average.

Figure 7: Simulation results in Throughput.

The comparisons of simulation results from

Figures 6 and 7 are summarized in Table 4, which

confirm our expectation that QPF can perform better

than PF. According to Table 4, the sequence of delay

time under QPF is Web (0.4) > VoD (0.23) > VoIP

(0.2) from the weighted relationship for application

service flows. It means that the VoIP service should

be handled first because the VoIP gets the minimum

delay. Also, the delay time of VoIP-qpf (0.2) is less

than VoIP-pf (0.25) which is enhanced by 20%.

In addition, from the simulation results of

Throughput, the QPF gains better efficiency than the

PF in average. Under different application services,

the QPF produces higher average Throughput than

the PF in which VoIP service can be enhanced by

8.19%, VoD service by 8.05% and Web service by

25.25%. Hence, QPF is adoptable to multimedia

application services, because of its advantages in

QPF SCHEDULING SCHEME FOR PERFORMANCE IMPROVEMENTS OF INTEGRATED MULTIMEDIA

APPLICATIONS OVER 3.5G NETWORK

terms of Delay and Throughput performance over

VoIP, VoD and Web application.

Table 4: Summarized simulation results.

KPI Service

PF QPF % of diff.

Delay

VoIP

0.25S 0.2S 20

VoD

0.26S 0.23S 11.53

Web

0.56S 0.4S 28.58

Throughput

VoIP

10.9Kbps 11.9 Kbps 8.19

VoD

5.89Kbps 6.51 Kbps 8.05

Web

3.49Kbps 4.67 Kbps 25.25

6 CONCLUSIONS

IMS through full IP is a good solution to provide

cross-platform integration service between different

regions on heterogeneous wired or wireless network.

However, in real-time transmission (e.g. VoIP), the

quality of data transferring may be affected by a

delay during network transmission. Service needs in

diverse applications flows should be considered

while transmitting signals over wired or wireless

network.

This paper proposes the QPF with different levels

of priority weights given as per diverse data type of

application service flows such as VoIP, VoD and

Web etc. We perform simulations with comparisons

on two KPIs of Delay and Throughput between QPF

and PF schemes, by using NS-2 software tool under

various multimedia application services, such as

VoIP, VoD and Web. The simulation results indicate

that QPF lowers 20% for VoIP delay and gains

8.19% more VoIP Throughput than PF. The QPF

performance for VoD delay can be enhanced by

11.53%, the Throughput is increased by 8.05%. Also,

the delay for Web of QPF can be improved by

28.58% and the Throughput is enhanced by 25.25%.

Consequently, the proposed QPF can provide better

services to corresponding flow levels upon different

applications over 3.5G network, which not only

improves the effectiveness and efficiency of

multimedia application services but also enhances the

QoS of overall network.

REFERENCES

3GPP, 2001. High Speed Downlink Packet Access

(HSDPA): Overall UTRAN Description, TR 25.855,

Release 5, V5.0.0.

3GPP, 2010. IP Multimedia Subsystem (IMS), TS 23.228,

Stage 2, Release 10, V10.1.0.

Ameigeiras, P., Wigard, J., Mogensen, P., 2004.

Performance of Packet Scheduling Methods with

Different Degree of Fairness in HSDPA. In Proc. of

IEEE Vehicular Technology Conference, 2, 860-864.

Aniba, G., Aissa, S., 2004. Adaptive Proportional Fairness

for Packet Scheduling in HSDPA. In Proc. of Global

Telecommunications Conference, 6, 4033-4037.

Dixit, S., Guo, Y., Antoniou, Z., 2002. Resource

Management and Quality of Service in Third-

generation Wireless Networks. IEEE Communications

Magazine, 39(2), 125-133.

Fathi, H., Chakraborty, S., Prasad, R., 2004. Optimization

of VoIP Session Setup Delay over Wireless Links

Using SIP, In Proc. of IEEE Communications

Conference, 6, 4092-4096.

Ghosh, A., Love, R., Ratausk, R., Xiao, W., 2004.

Performance of 3GPP High Speed Downlink Packet

Access (HSDPA). In Proc. of IEEE Vehicular

Technology Conference, 5, 3359–3363.

Holma, H., Toskala, A., 2006. HSDPA/HSUPA for

UMTS, High Speed Radio Access for Mobil

Communications. John Wiley and Sons, Ltd. 1

edition.

Jalali, A., Padovani, R., Pankaj, R., 2000. Data

Throughput of CDMA-HDR a High Efficiency-High

Data Rate Personal Communication Wireress System.

In Proc. of IEEE Vehicular Technology Conference, 3,

1854-1858.

Krause, J., 2002. ”High Speed Downlink Packet Access:

Iub/Iur Protocol”, 3GPP Specification Release 5.

Kara, N., Planat, V., 2007. Performance Analysis of IP

Multimedia Services over HSDPA Mobile Networks.

In Proc. of IP Multimedia Subsystem Architecture and

Applications, 1-5.

Li, Wei., Aaron Gulliver, T., 2009. Performance of

Enhanced Proportional Fair Scheduling in HSDPA for

Multimedia Service Streaming, In Proc. of

Communications, Computers and Signal Processing,

314-318.

Shreevastav, R., McGoldrick, C., Huggard, M., 2009.

Delivering Improved QoS and Cell Throughput in

UMTS Based HSDPA Network”, In Proc. of World of

Wireless, Mobile and Multimedia Networks, 1-9.

ICEIS 2011 - 13th International Conference on Enterprise Information Systems