Monitoring and improving performance in IMS networks

raggedsquadNetworking and Communications

Oct 30, 2013 (3 years and 9 months ago)

218 views

Moni
toring and improving performance
in
IMS networks







Innova
tive
Communications
Technologies &
Entrepreneurship

8
t h

S e m e s t e r

5/2 4/2 0 1 2

S t u d e n t
:

Alexandros Fragkopoulos

Group Number:
12
GR
890


Supervisors:

Rasmus Hjorth Nielsen

Neeli R. Prasad



1

Alexandros Fragkopoulos





ICTE 8




Abstract


During the last decades the development and improvement of telecommunications is
massive. If one considers how long ago

the telephone was

a privilege
that only a few
possessed
, then one can see the big onward steps technology has made.
In this rapid rhythm
of technological advances, people demanded to be connected more

and more time

until
they reached the point that

they w
ant anywhere and anytime

full access into the digital
world. It either serves their needs for work or entertainment. Looking in this direction,
research has been made the last years on the development of efficient usage of the internet
services via smartph
ones or any other portable device

and through any type of network
. An
important issue that has to be solved and clarified is how does someone utilize

same
services through

all these different technologies (old and new, complex and simple, wired
and wireless oriented)? And if one manages to do that, then who is benefited from that? All
these questions bring more and more issues on the surface. In an effort to address all the
se
issues the IMS framework has been created. Although it is relatively still n
ew as a
framework, numerous benefits have been discovered

through

research for

this network.
I
MS
networks help to interconnect devices through different tec
hnologies and still p
rovide

the
same quality and number of services. In this way users are satisfied by the operators and
operators find the opportunity to setup a unified charging policy (through the IMS network)
that will help them increase their profits. This project though
, is being focused on the
performance of these networks from the point o
f improving them but also defining

what are
the most important KPI’s that depict weaknesses in performance issues. Concerning the part
of improvement, some ideas from two pap
ers are be
ing presented and

an approach of them
is being implemented. In the

end of this project
some results

from simulations are derived
and discussed
. Last but not least is

a small reference to the future technologies

such LTE and
LTE
-
A
.







2

Alexandros Fragkopoulos





ICTE 8




Table of Contents


Abstract

................................
................................
................................
...........................
1

List of Figures

................................
................................
................................
...................
3

List of Tables

................................
................................
................................
....................
3

List of Abbreviations

................................
................................
................................
.........
3

Chapter
1


Introduction

................................
................................
................................
...
5

Chapter 2
-

IMS
................................
................................
................................
.................
7

2.1 Why IMS?

................................
................................
................................
...............
7

2.2 3GPP: Requirements and Standards

................................
................................
.........
7

2.3 IMS Architecture

................................
................................
................................
.....
9

2.4 Entities inside
an IMS
................................
................................
............................

10

Chapter 3


Analysis
................................
................................
................................
.......

15

3.1 State of the art

................................
................................
................................
.....

15

3.2 Performance in IMS services

................................
................................
.................

16

3.3 KPIs for the IMS network

................................
................................
......................

17

3.3.1
Accessibility KPIs

................................
................................
............................

17

3.3.2 Retainability and Utilization KPIs

................................
................................
.....

19

Chapter 4


Design
................................
................................
................................
.........

20

4.1 First use case
................................
................................
................................
........

20

4.2 Second use case

................................
................................
................................
...

21

4.3
Improving performance in IMS networks

................................
...............................

22

Chapter 5


Implementation

................................
................................
..........................

23

5.1 Description of the algorithm

................................
................................
.................

23

5.2 Figures and Results
................................
................................
...............................

25

5.3 Results’ Analysis

................................
................................
................................
...

28

Chapter 6


Conclusions

................................
................................
................................
.

30

6.1 Future Vision in Communication Technologies

................................
.......................

30

Bibliography

................................
................................
................................
..................

31




3

Alexandros Fragkopoulos





ICTE 8




L
ist of Figures


Figure 1
-

Layered Architecture of the IMS [2]

................................
................................
....
9

Figure 2


Successful
IMS Voice Session Establishment [1]

................................
...............

13

Figure 3


Plots from First Case Scenario

................................
................................
.....

25

Figure 4
-

Plots from Second Case Scenario

................................
................................
..

26

Fi
gure 5
-

Plots from Third Case Scenario
................................
................................
......

27

Figure 6
-

Plots from Fourth Case Scenario

................................
................................
...

28

Figure 7
-

Ranking of the Best Case Scenario
................................
................................

29

List of
Tables


Table 1


Acquisition of data from the simulations
................................
........................

28


List of Abbreviations


PSTN



Public Switched Telephone Network

IP



Internet Protocol

DSL



Digital Subscriber Line

3G



3
rd

Generation

4G



4
th

Generation

IMS



IP Multimedia Subsystem

3GPP



3
rd

Generation Partnership Project

VoIP



Voice over IP

QoS



Quality of Service

QoE



Quality of Experience

GPRS



General Packet Radio Service

LAN



Local Area Network

WCDMA


Wideband Code Division Multiple Access

WLAN



Wireless Local Area Networ
k

WiMAX



Worldwide Interoperability for Microwave Access

OFDM



Orthogonal

Frequency Division Multiplex




Flash
-
OFDM


Fast Low
-
latency Access with Seamless Handoff OFDM

LTE



Long Term Evolution

CSCF



Call Session Control Function

HSS



Home Subscriber
Server

P
-
CSCF



Proxy
-

Call Session Control Function

I
-
CSCF



Interrogating
-

Call Session Control Function

S
-
CSCF



Serving
-

Call Session Control Function

E
-
CSCF



Emergency
-
Call Session Control Function

AS



Application Server

4

Alexandros Fragkopoulos





ICTE 8




SIP



Session Initiation Pro
tocol

SLF



Subscriber Location

Function

MGCF



Media Gateway Control Function

BGCF



Breakout Gateway Control Function

DL



Download

UL



Upload

SISO



Single Input Single Output

MIMO



Multiple Input
,

Multiple Output

CS



Circuit Switched

CSN



Circuit
Switched Network

NACF



Network Attachment Control Function

IP
-
CAN



IP Connectivity Access Network

UE



User Equipment

UA



User Agent

CLF



Connectivity Session

Location and Repository Function

CN



Core Network

RAN



Radio Access Network



5

Alexandros Fragkopoulos





ICTE 8




Chapter 1


Introduction


Telecommunications is a very valuable and indispensable part of people’s lives. In every
aspect of everyday life there is an increasing need to communicate with people that are far
from someone. This need started in the ancient times for diff
erent reasons and purposes
than

nowadays. Since then it developed

and today we’ve reached to a point that
telecommunications is a very important part of our work, lives and also free time activities.
Since telecommunications is so important in various diff
erent levels, there has being great
effort to combine different types of telecommunications. In reality combining different types
of telecommunications is much more complicated t
han it looks like. There are old

(PSTN)

and new

(IP networks)

technologies tha
t have to be combined. Another factor should also be
met and that is the combination of wired (DSL) and wireless networks (3g/4g). An effort in
combining all these networks, technologies, mediums etc. under one architecture is called
IMS (IP Multimedia Sub
system) networks.
In the evolving world of telecommunications the
need for creating converged networks has prevailed and the IMS networks pointing to this
direction. Altho
ugh th
ey already operate in many places of the world

many things still
remain unsolve
d.

T
his project will give an insight on how IMS networks function, which are their features

but
also which are their
advantages and
dis
advantages in

combining different technologies.
Furthermore, there will be an analysis on how does someone monitor perfor
mance on these
networks and which are the challenges. Significant

key factor in performance of the
networks is securit
y but this project will not get into security issues.



Motivation

IMS networks seem to be the next big thing in the market of
telecommunications from the
point of combining different technologies and exchange content between them, making our
communication experience richer and the technologies interoperable. Since IMS are in the
center of attention, industry is pushing through R&
D (research and development) for
improvement of these networks and defining some aspects of them that are not yet so well
defined, such as performance or security. By allowing this act as an incentive, this project
focuses on matters of performance.

Perfor
mance is quite important in general but
specifically in IMS networks, performance is really important since it needs to:

1.

B
e well define
d. Which are the performance characteristics that affect
an IMS network and why.

2.

How could one improve the performance on

a framework that uses
different technologies?

By monitoring and better understanding of the ideas behind performance and control, one
would be able to set the best parameters or make the most optimum alterations in order to
6

Alexandros Fragkopoulos





ICTE 8




have the best possible results,

and thus efficiency in a network that has IMS functionality.

Thus, this project is trying to address some issues in two entities of the IMS networks, the
Proxy
-
CSCF and the Server
-
CSCF. There is still
space

for

improvement there and by
investigating some alternatives on how the user could get connected in these two entities;
one could derive useful information concerning the general improvement of the IMS.



Problem Formulation

Performance in IMS networks is of
vital importance as the wireless networks and services
becoming more demanding and QoS needs to be increased. The idea is that in this project
there will be a description of the main KPIs (Key Performance Indicators) and then there will
be an effort to det
ermine how some changes in the parameters of the IMS or some
extra
attention in the aforementioned entities

could increase their efficiency and performance.

This will be done by the introduction of different cases
for the same problem and then there

will

b
e

a
discuss
ion

on which approach fitted better and why.





7

Alexandros Fragkopoulos





ICTE 8




Chapter 2
-

IMS


In this chapter there will be an analytic
al

description of what is an IMS


through
investigation

of why do we need it

, which are its requirements and how does its entities
function. There will also be examples of how the services are provided

to the end users

in
certain cases.


2.1 Why IMS?


Cellular systems have been around for a couple of
decades. Circuit switched networks have
made a big impact


and still do


in cellular systems with the use of voice service. It is
,

until
today
,

the most popular and reliable deployment for cellular networks. In the cellular
networks there is an effort to bring all services together under one roof. In order to
accomplish that,
cellular systems have been using

for

some years

now,

IP in fixed parts
of
the network, whereas as air interface not that much. Since two to three years there has
been a great effort to try and include real time services over the air and use IP also there
until the end user and not only in the radio access and core network. In
troduction of real
-
time services in cellular systems is linked with VoIP (Voice over IP) technology. So, the near
-
future vision is to have packet switched networks and end to end IP usage regardless if it is
fixed or mobile systems. Migration from the curr
ent network is quite difficult but it can be
phased and cut down into parts.

Packet switched networks are deployed to cover the need for a more interactive form of
communication via any type of device and between any technology. It is possible through
them

to exchange voice, data, video and messaging and in that way

communication is also
enriched
. In order to enable simultaneous usage
of

circuit

switched

networks together with
packet switched networks and to have an equivalent quality of packet switched net
works as
with circuit
switched

an

architecture

is being specified by 3GPP and is called IMS

(IP
Multimedia Services)
. IMS

includes this functionality to enable

these different types of
networks to co
-
exist
.

In IP multimedia services there are some parts o
f the protocol stack that can be reused such
as the addressing and routing, whereas other, like parts from the application logic, differ in
order for the service to be unique. Thus, there is a common base in IMS networks and
services are interoperable with

others

[1]
.


2.2 3GPP: Requirements and Standards


In order for the IMS to be

able
to

cover a wide spectrum of technologies it should contain
some requirements that an IP multimedia application should fulfill. By being standardized it
functions as

a start
ing

point for

a common basis for

manufactures,
operators and vendors.
These requiremen
ts are mentioned below:

8

Alexandros Fragkopoulos





ICTE 8






Potential for QoS negotiation during a session but also at the beginning of it.



End
-
to
-
end voice quality the same or greater than circuit switched telephony
services.



Support of roaming and negotiation for QoS and service capabiliti
es between service
operators.



Reassurance of interoperability with default media types between all the services.
But also support for other media types.



Possibility to have several IP multimedia
applications

within each IP multimedia
session.



Same level of

privacy, security and authentication protection with GPRS and circuit
switched services.



Access independence support. So the users could be able to connect through
different technologies (such as: GPRS, fixed lines, LAN, WCDMA (3G))



Potential to support
applications that have been developed outside 3GPP
community.

All of the above requirements bring a unified structure for all different applications and
benefit users, vendors and operators in different ways each of them.



Users benefit from rich multimedia

applications and rich services provided in within.
They are also benefit from the high level security and integrity while inserting their
personal data. And last but not least, users benefit also from the possibility of using
one device and passing from o
ne technology to the other and still being able to use
their services seamlessly but also being able to use their services through different
terminals.



Operators benefits from common authentication and authorization mechanisms,
service control and fraud ma
nagement, charging mechanisms but also legacy
telephony interworking. One last thing that an operator benefits from is that IMS
helps to improve efficiency by providing information concerning radio bearer
establishment process and services. And indeed this

improves the efficiency


for
the operator


as it aids

to wisely select a radio bearer and make use of header
compression as a performance booster.



Vendors benefit from the introduction of one technology platform for various
services because it works as
an incentive in improving infra
structure and terminal
devices

[1]
.



9

Alexandros Fragkopoulos





ICTE 8




2.3 IMS Architecture


IMS basically consists of three layers and in these layers there is communication for two
different reasons: signaling and data.
Some components exchange only signaling but others
data and signaling.


Figure
1

-

Layered Architecture of
the IMS
[2]


This happens because only at the “lowest” level/layer of bit exchange
,

there is the need to
exchange data between users. The aforementioned layers are the following

(from bottom to
top)
:

1.

Media/transport

layer

2.

Control/Signaling

layer

3.

Service/Application

layer


Each layer aids in a specific way so that a communication proce
ss between different
technologies and architectures could be achieved successfully.



The media/transport layer includes different types of access technologies such as
WLAN, 3G, xDSL, GPRS
, WiMax
,
Flash
-
OFDM

and LTE, that users can access in order
10

Alexandros Fragkopoulos





ICTE 8




to be
interconnected to the network. In this layer except signaling there is also data
exchange, as mentioned above.



The Control/Signaling layer includes the most important components of IMS which
are all the **CF (Control Functions) and the HSS (Home Subscriber

Server). The CSCF
(Control Session Control Function) are the following: Proxy
-
CSCF, Interrogating
-
CSCF
, Serving
-
CSCF and Emergency
-
CSCF. The Control functions will be explained later. In
this layer the SIP signals are being processed and routed to the de
sired destination.



The Service/Application layer includes various AS (Application

Servers) that host and
execute a wide range of services through the IMS architecture

[2]
.


2.4 Entities inside an IMS


In this section there will be an insight to the entities of the IMS and a description of their
usability. There will also be a reference to SIP and DIAMETER protocols, as they are the most
important protocols in this architectural framework. And since SIP
is the “messenger” inside
most entities of the IMS, it will

be

describe
d

before any entity.


SIP in IMS

SIP (Session Initiation Protocol) is the most important signaling protocol of the IMS. SIP is
used for

arranging sessions and establishes them but also manage
s and controls

them.
Basically, in th
e Control and Signaling layers are the places

where this protocol

is

mostly
used. The SIP protocol is based in requests and acknowledgements or rejections.

These
m
essages are sent in the form of a number and a word summarizes the function of the
message. These numbers are three digits
long

and the first digit gives information about the
type of the message. Below

are presented

the six different types
of messages
:

1x
x: Provisional Messages

2xx: Successful answers

3xx: Redirection Answers

4xx: Method Failures

5xx: Server Failures

6xx: Global Failures
[3]

All these messages are being exchanged between the entities in an IMS.
The complete
process on how does SIP exactly functions will be explained

in detail

below.



11

Alexandros Fragkopoulos





ICTE 8




CSCF

Call/Session Control Functions are components that make use of SIP signaling and help in
various cases


such as establishment, monitoring and controlling sessions


and ar
e
consisted of P
-
CSCF, I
-
CSCF, S
-
CSCF, E
-
CSCF. Generally, CSCF is responsible for the following
matters:



Keep track of session status



Querying HSS for authentication information concerning users



Establishment and allocation of the resources and routing of
the information
through the correct nodes for a successful connection

A more detailed description of the components
of the CSCF is presented below

[4]
.


P
-
CSCF

The Proxy
-
CSCF is the component that makes the direct communication

with the user. Any
IMS
requests from the user are routed to

a

P
-
CSCF

node, through a RAN (Radio Access
Network) or a WLAN or in general any other type of network

and from

the

P
-
CSCF

node

to

an

S
-
CSCF

node
. P
-
CSCF features are SIP compression (SigComp
(Signaling Compression)),
interaction with PCRF and establishment a mutual authenticated communication with t
he
user with the help of IPsec

[5]
.
SigComp is used because if the available bandwidth is low,
then the establishment
of the connection will take a long time to be completed. In order to
avoid that,

a

compression

method

is used

[1]
.


I
-
CSCF

The Interrogating
-
CSCF is responsible for querying the HSS to assign a
n

S
-
CSCF

node to the
user
that have communicated with the P
-
CSCF, and by querying the HSS, then the HSS
assigns an S
-
CSCF to the user/subscriber. The I
-
CSCF together with P
-
CSCF were also enrolled
to hide the rest of the IMS network from the users and making a hiding topology but t
hat
happened until

the 7
th

release of

3GPP. Thereafter
,

these two elements stopped having this
function and became part of the IBCF (Interconnection Border Control Function)

[5]
,

[6]
.


S
-
CSCF

The responsibilities of the Serving
-
CSCF are about: maintaining sessions, decisions on how to
route data, storage of service profiles and handling for the SIP registrations coming from the
subscriber. The S
-
CSCF checks in the HSS whether the user is authen
ticated to perform a
certain registration
for a certain service. After the HSS’s

approval
,

the S
-
CSCF continu
es
to
monitor

the registration

[5]
.

Another important process

that

is used by the S
-
CSCF in order
to provide routing services and support

the

connectivity with older technologies such as
PSTN or ISDN is ENUM (E.164 Number

Mapping). ENUM is used

to interconnect the
12

Alexandros Fragkopoulos





ICTE 8




telephone number based systems (such as PSTN) to the URI’
s based system (such as SIP
protocol). This function is managed by ENUM with the translation of telephone numbers
into URI’s.
‘E’ in

ENUM
stands for

E.164 ITU
-
T standard
for international numbering

where
all globally
-
reachable telephone numbers are registe
red and organized

[7]
.


E
-
CSCF

The Emergency
-
CSCF is a newer component in the control functions and is responsible

for
controlling the request of

an emergency call. The subscriber is calling an emergency number
and the SIP request is being directed directly to the E
-
CSCF through the SGSN (Serving GPRS
Support Node). Information concerning the user’s location is also important and that is the
respon
sibility of HSS and GMLC (Gateway Mobile Location Center) collaborating to locate
the SGSN and send the information to the E
-
CSCF and then to determine which is the close
st

PSAP (Public Safety Answeri
ng Point)

[8]
.


MGCF &

BGCF

In order to connect to a circuit switched network through the IMS platform, all the SIP
messages from the Control Functions should be altered to comply with the signaling
protocols in this type of network. Because of that, there are two functions that imp
lement
the translation, BGCF (Breakout Gateway Control Function) and MGCF (
Media Gateway
Control Function).

In order for t
he messages to be translated

MGCF uses a H.248
-
based Mn interface. With the
aid of this interface, the RTP (Real Time Protocol)
-
based
data/media, are translated in the
media format which is acceptable by the circuit switched network. The interwork with the
circuit switched network is accomplished by the use, from both sides, of a circuit switched
signaling over IP.

Finally, the control f
unction of the BGCF is actually choosing which MGCF will handle the
session, whether it is in the same domain or in

a different one, where, in the latter

case, it
directs the

SIP

req
uest to the desired BGCF

[1]
.


HSS
&

SLF databases

HSS and SLF are the databases used in IMS. HSS is the database that includes all the
subscribers’ data such as: identities, access parameters, service
-
enabling information and
registration information. The SLF (
Subscriber Location Function
)
on the other hand is the
responsible database that gives information to the I
-
CSCF and S
-
CSCF for which HSS has the
information of a specific user,

even

in case there is more than one

public user identities
. All
these information that are being sent from a
nd to, HSS and SLF, make use of the DIAMETER
protocol, which is a networking protocol for AAA (Authentica
tion Authorization Accounting)

[9]
.


13

Alexandros Fragkopoulos





ICTE 8




SIP’s Functionality

In

Figure
2


it is presented how SIP signaling will function in order to establish a session.


Figure
2



Successful

IMS Voice

Session
Establishment

[1]

14

Alexandros Fragkopoulos





ICTE 8




Initially, there is a SIP Invite request to the P
-
CSCF entity. This entity sends the request to the
S
-
CSCF that the HSS assigned the user to. Then, the request is being forwar
ded to the I
-
CSCF
where
this entity queri
es HSS for information about the other S
-
CSCF. The query i
s

sent to
the HSS with the aid of DIAMETER protocol. After a successful reply from the HSS, the I
-
CSCF
connects with the other S
-
CSCF that makes an evaluation of the user’s filters and controls
whet
her it needs an AS invocation or not. Afterwards, the request is being forwarded to the
user through the P
-
CSCF that is assigned for it. After the process is completed and before the
establishment of the session,

a provisional response is

sent and at that

time both terminals
try to reserve resource from their access network. In this way there is an effort to reassure
QoS during the initiation of the session. After the QoS is achieved the terminal starts ringing
and the session is established. The example s
hown above describes a successful

establishment of a Voice session

scenario between two terminals through the IMS network

[1]
.





15

Alexandros Fragkopoulos





ICTE 8




C
hapter
3


Analysis


In this chapter there will be a reference to state of the art work

which is happening

in the
field of IMS. There will also be a detailed explanation of all important KPIs

that play an
important role in improving IMS networks and how to estimate the improvement accurately.


3.1

State of the art


In the area of the IMS
,

and specifically
their performance, various papers

have been
published. These papers focus on different aspects

and from different angles

of the
performance

indicators

of an

IMS netw
ork. Some
papers

focus specifically in the internal
performance of the IM
S
(e.g. the cooperation between

the CSCF entities)

and some others
to the collaboration of
an IMS network with RANs

that utilize

it.

In

[10]

the case was that
they observed several RANs and
compared them under the same
basis.
By

plotting some important time parameters

for sessions, they gave an idea about
faster and slower combinations of different RANs. According to their paper, they
investigated

four different times in sessions: signaling request time, signaling reply time,
sig
naling release time and
total session setup and release time. By doing these
computations, they came to the conclusion that when the end users use GSM network to
connect to each other, the
y have the highest delays in

all four measurements. On the other
han
d, users that use WLAN from both sides have the least
delays

in all four categories. So,
according to this paper, the best performance in heterogeneous access networks is the
WLAN
-
IMS
-
WLAN connection.

As a conclusion it is stated that various factors affec
t these
times such as: which technology ones uses, different access nodes, capacity of the IMS core
network and
available bandwidth of RANs.

Another paper

[5]

goes

more into detail by
comparing the performance of the IMS networ
ks
to the performance of the SIP

based

networks. Under the same workload both networks
have been tested on messaging delays and session initiation delays.

As messaging
procedure, it is stated the delivery of the message and the acknowledgement for its
successful delivery. As the session initiation procedure, it is determined the time from the
first INVITE message until the acknowledgement from the called pa
rty.

There were three
workloads: 200,400 and 600scenarios/sec. The aim was to observe in which network the
time needed to either initiate a session or send a message, was higher. Some results from
this paper project that both networks performed quite well
but the SIP

based

network was
faster in both cases and under all workloads. The users that were exchanging these
messages o
r initiating these calls

had different statuses and they
could be registered or not
-

available

or executing some other scenario. The

selection of their statuses was based on a
Poisson distribution.

One last thing that it is included in these results is the addition of a
probability that 50%, 90%, 95% or 99% of the service was handled smoothly. So one could
see where the time was increa
sed and under which circumstances (what amount of
workload).

The last work that is presented here is entirely based on internal performance of the IMS
core network. According to

[11]

there have been several tests regarding diff
erent issues in
an IMS core network. The first test determined some performance factors by increasing the
number of simultaneous calls/sec
. It was observed that as long the server could handle the
16

Alexandros Fragkopoulos





ICTE 8




routing of more calls everything were steadily increasing.
The number of calls/sec was
increasing and the processor load was increasing
too; until the point that the number of
simultaneous calls/sec reached to the value of 150 and the server rebooted due to
overloading.
Throughout the test, the memory utilization
didn’t exceed 65%.

The second
test was more focused into security and robustness matters by sending malformed INVITE
requests.
The m
essages were sent either to the UA
(User Agent) or the SIP proxy server.
These messages were containing exceptional
elements; that is
,

data that could arise
undesired behavior to the receiving point as the device or process could crash, or needed to
be rebooted manually or consume a considerable amount of memory and/or CPU for a fair
or infinite amount of time. If one o
f the above actions takes place
, the test fails. The third
and last test that was introduced in this paper evaluated the behavior and the response of
the system when sending SIP requests. The test cases that were used are introduced in the
ETSI TS 102 027
-
2

V4.1.1. The test cases included

messaging, call control and querying
capabilities series. The conclusions that were
drawn

showed that all three tests

passed

successfully
. The first one determined the maximum amount of simultaneous routable calls
and it c
ould be used to benchmark the capacity of different vendors. The second one passed
the test, as there was no failure to the devices or the services due to malformed messages.
The third test finished also successfully and the responses were correct accordin
g to the RFC
3161 standard.


3.2

Performance in IMS services


Performance is an important factor concerning the efficiency of an entity. When there were
only circuit switched networks, measuring performance was easy. For voice services the
performance was
measured as the highest amount of satisfied users that the system could
support. Nowadays, cellular system
s support also VoIP technology via

IMS connectivity. In
VoIP, things
are not that clear, as to how one

evaluate
s

a performance of a network that is
based not only in voice services but also in video and other components. A point of such an
inquiry is the interaction between different forms of multimedia, such as audio and video

[1]
.


Service Performance Requirements

Services have some characteristics in order to be evaluated concerning their performance.
Some of them are: service availability, retainability and quality. A circuit switched network
has a minimum downtime of 0.0001% a
nd some similar standards for speech quality and
mouth
-
to
-
ear delay. VoIP standards are not as high as in the circuit switched networks.

Users perceive two characteristics as far as concerning performance through VoIP
telephony. The first characteristic t
hat affects users’ experience is


individually


every form
of multimedia. For example, for video sessions, the codec that is used has a large impact on
the quality and performance of the service and for voice sessions the speech quality has
more aspects,

such as: sampling rate, frame loss etc. On the other hand, the quality of the
services, individually, is not enough because of the potential end
-
to
-
end delay of the media.
17

Alexandros Fragkopoulos





ICTE 8




So the second characteristic is the combination of them, for a high

quality end
-
to
-
end
experience

[1]
.


3.3 KPI
s

for

the

IMS

network


According to 3GPP and its latest update

TS 32.454 V11.0.0 (2011
-
12) on

Key Performance
Indicators (KPIs) for the IMS networks
, they

are divided to three major categories and these
include, the Accessibility KPIs from network and user perspective, the Retainability KPIs from
network perspective as well as the Utilization KPIs from network perspective

[12]
.


3.3.1
Accessibility KPIs

The accessib
ility KPIs are divided into various

subcategories for a better understanding and
defining of the performance of the IMS networks, both from user and network perspective.
The KPIs for the accessibility of the IMS networ
k are listed and presented below.



Initial Registration Success rate of the Serving
-
CSCF



Session Setup


mean


Time



Session Establishment Success Rate



Third Party Registration Success Rate



Re
-
registration Success Rate for the S
-
CSCF



Session Setup


mean


Time (regarding the messages originate from an IMS)



Session Setup


mean


Time (regarding the messages
originate from
a

CS
N
)



Immediate Messaging Success Rate



Session Es
tablishment Network Success Rate

[12]


Initial
Registration Success rate of the Serving
-
CSCF:

This KPI evaluates the success rate concerning the number of successful registrations to the
S
-
CSCF over the number of attempted registrations. This evaluation aids to control the
accessibility of the network

[12]
.


Session Setup


mean


Time

This KPI acquires the mean setup time for a session. It is very important to know if the setup
times are low or high, both from users’ perspective and satisfaction but also from network
transa
ction performance

[12]
.



18

Alexandros Fragkopoulos





ICTE 8




Session Establishment Success Rate

This KPI evaluates the session establishment success rate by giving two results from two
fractions representing two different perspectives. One from the originating
side and one
from the terminating side of the session establishment. These two fractions are the
successful session establishments over the attempted session establishments. This KPI
differentiates between session successes from the originating and the ter
minating side, in
order to have the real number of success rate. It also includes users’ behavior as no attribute
is excluded

[12]
.


Third Party Registration Success Rate

This KPI calculates a very vital feature of the IMS netw
orks and that is the success rate
regarding registration to third parties. That is the information provided by the S
-
CSCF to the
AS

(Application Servers) as to whether a user is registered or not. The KPI is calculated by the
successful 3
rd

party registrations over

the attempted ones

[12]
.


Re
-
registration Success Rate for the S
-
CSCF

In this KPI it is investigated whether the success rate of re
-
registration is high or not. This is
calculated by the success re
-
registrations over the attempted ones. The re
-
registration is
useful for
some

reasons; it

can either

aid to inform the network o
f a change into the
registration status or just refresh the existing status or to inform the network about a
change in the capabilities
of the UE or possibly to inform for a change in the IP
-
CAN

(IP
-

Connectivity Access Network)

[12]
.


Session Setup


mean


Time (regarding the messages originate from an IMS and CS)

Another KPI is the mean time regarding Sessions’ Setup from
the IMS CN (Core

Network) and
the mean time regarding the Sessions’ Setup from the CS point of view. I
n both cases the
results are helpful to IMS and to other network operators as they can control whether the
performance issues originating from the IMS endpoint or the CS side

[12]
.



Immediate Messaging Success Rate

In
immediate messaging service the content could vary as there could be any type of
multimedia. The success rate affects users’ experience and is expected to be high. It is
calculated by the number of successful immediate messaging procedures over the
attempt
ed immediate messaging procedures

[12]
.

19

Alexandros Fragkopoulos





ICTE 8





Session Establishment Network Success Rate

This KPI gives the performance of the IMS based on the successful session establishments
together with the number of failed session establishme
nts due to users’ behavior over the
number of attempted session establishments. This KPI gives an insight to the actual success
rate of the sessions that have been established and therefore a more accurate evaluation of
the performance of the network, conc
erning this aspect. In this KPI there is again a different
calculation for the originating and the terminating side of the session establishments

[12]
.



3.3.2
Retainability and Utilization KPIs

The KPIs above are responsible
for addressing problems or malfunctions in the network or
just evaluation of the normal behavior by controlling accessibility to the network. But there
are also other forms of performance matters that can come up. Below the last two KPIs will
be discussed,

concerning retainability of a session and utilization of an IMS network. So
these aspects are:



Call Drop Rate of IMS Sessions



Mean Session Utilization

[12]


Call Drop Rate of IMS Sessions

This KPI helps for the evaluation of t
he retainability of the sessions. In order to calculate this
rate, the fraction would be the amount of dropped sessions over the number of successful
ones. This is also an important key indicator but doesn’t give a lot of insight as to where
could be the p
roblem. It indicates perfo
rmance matters that

could be caused by the IMS
network or the user side

[12]
.

Mean Session Utilization

This KPI tries to address the utilization of the network by calculating the mean number of
simultaneously online and answered sessions over the capacity of the network. This
indicator reflects the relation between the size of the network and the utilization of it. So, in
case this number is low, this gives an indicator that the network is utiliz
ed up to a good level

[12]
.


20

Alexandros Fragkopoulos





ICTE 8




Chapter 4


Design


All the
KPI’s

in the chapter above test

the performance of the IMS networks from different
aspects and from different technologies that cooperate with the IMS networks. In this
chapter the project will

move one step further and

provide an insight to what exactly is
happening when the net
work re
aches its limitations. F
rom that point and after
,

its entities
become unstable and the whole system can’t provide services until it reboots or until it
overcomes the overload of the processor.

T
here will

also

be an analysis on how the sessions
are directed

into the IMS

when balancing a load of calls

and what is happening while trying
to establish the nearest P
-
CSCF in a net
work
. Both cases

mentioned

will be repr
esented by

two

scenarios.


4.1 First use case


In the case that there is an increased amount of s
essions that needed to be setup, then the
entire load is being directed from the P
-
CSCFs to the S
-
CSCFs by having the I
-
CSCFs choose


via HSS information


which S
-
CSCF will make this setup for its session.


Root cause of the Problem

The large number of s
essions is not a problem for an adequate number of S
-
CSCFs. Though,
if the amount of traffic is not
well balanced then problems might

occur as one of the Serving
nodes will be overloaded with requests or established sessions and will cause a restart to it.

As a result all the sessions of data, video or audio will be lost and the performance of the
networ
k will drop significantly. So,

S
-
CSCF is playing an important role in the good
performance of the whole IMS architecture.


So
lution

As stated
in
[13]

I
-
CSCF is responsible for assigning the users to an S
-
CSCF but the control for
an overloaded S
-
CSCF comes afterwards. That means that a I
-
CSCF directs a user to the S
-
CSCF and if the S
-
CSCF

node

become overloaded then the
I
-
CSCF starts de
-
register users
from there and registers them to another S
-
CSCF

node
. The problem that derives from this
action is
low

performance
and

a possibility that the user will be assigned again at the same
S
-
CSCF

node

as it will not seem to be over
loaded after a number of de
-
registrations.
So, in
[13]

there is a proposal that supports the use of a load balancing method and that

this
function should be handled by

I
-
CSCF before even registering the user to a S
-
CSCF. This
m
ethod does not add any new entities or any significant message signaling overhead. The
change is that the selection of the S
-
CSCF is done basically in the I
-
CSCF by knowing
beforehand which S
-
CSCF has available capacity to support more load.

This is done b
y SIP specific event notifications which will be sent from the S
-
CSCF to the I
-
CSCF and inform every time about the load level of the entity. The criteria that
should be
21

Alexandros Fragkopoulos





ICTE 8




fulfilled are

that the S
-
CSCF has free capacity to serve the user and that it has the
required
capabilities to do it. They propose that the notification messages are sent only every time
there is a change in the load of the S
-
CSCF and the content of these messages will be only
the identifier of the S
-
CSCF and the value that indicates the am
ount of load in it. That
message has the size of 512bytes.

In this way the I
-
CSCF knows every moment which S
-
CSCF
has available
slots

to accept a new user for registration. In the case that all available S
-
CSCFs
are full, a 5xx SIP message is
sent, informs

that the server has

failed.

Having an impr
oved load balancing system provides

a positive effect in nearly all KPIs which
are stated in the TS 32.454 V11.0.0 (2011
-
12). These KPIs concern the Accessibility,
Retainability and Utilization of the IMS network.

Generally, if one improves how load is
balanced in the S
-
CSCF entity then all the
mean times are decreased and

reliability and
durability of the whole architecture is improved.


4.2
Second
use case


In the

current case

there is the need for a UE to regist
er, the first step is the connection of
the UE to a P
-
CSCF

from which the UE will enter the IMS network
.


Root cause of the Problem

Although this scenario does not imply any fail of the system or some entity individually, it
presents the need of a new
approach to get connected more intelligently and smoothly to a
P
-
CSCF.


Solution

In
[14]

they propose a model based on two individual things. Initially, the idea is that every
P
-
CSCF will be characterized by two factors: firstl
y a geographical location (coordinates) and
secondly four pointers stating where the P
-
CSCF is located by having NE, SE, SW, and NW as
signs. Then, this idea will be used by a near neighbor range search in two dimensional quad
tree, algorithm and the UE wi
ll get connected to the correct P
-
CSCF regarding its
geographical location. This paper is more focused on th
e condition that the UE is moving

along visited networks.

The quad tree algorithm is functioning by dividing the two dimensional space into four
quadrants with the initial point and then subdividing into four sub
-
quadrants every time
there is a new point noted. In every quadrant there is a characterization with N
E, SE, SW and
NW and it is helping in keeping a good architecture over the created network. This algorithm
proved to be useful because by stating which node (with the aid of coordinates) and which
quadrant, it actually gives the right information about the

location of the UE.

22

Alexandros Fragkopoulos





ICTE 8




CLF hold the quad tree

information with all the P
-
CSCF nodes

and the creation of four
quadrants for ea
ch
P
-
CSCF

node
. P
-
CSCF will provide the CLF, through e2 interface, with local
information. Executing the algorithm determine which
nodes are actually nearest to the UE
and

then check which
one of these has an available capacity. The information regarding the
active sessions are stored in CLF together with the information on the relay agent that is
used for each session. The informatio
n on the relay agent
is

sent via the a2 interface from
NACF. When the right P
-
CSCF is found, then CLF will inform NACF with a Bind
Acknowledgment. This message also contains the identity of the P
-
CSCF that serves the
session.


4.3
Improving performance in
IMS networks


Above t
here was a detailed reference in two improvements about

two different
entities

of
the IMS network. Both ideas contribute in two things and these are performance and
availability.

The first improvement is targeted in the part where all
the decisions are taken and a big part
of the intelligence of the IMS network is located and that is the S
-
CSCF. Whereas, the second
improvement is located in an

“entry”

entity that connects users with the rest of the IMS
architecture and this is the P
-
CSC
F. The c
ombination of them can be

applied since there is no
conflict between them
. The improvement in the P
-
CSCF is using a modified algorithm for
discovery of the P
-
CSCF and some changes in the way DHCP messages are utilized
. The
improvement though in the

other scenario adds a small overhead to the signaling between
the I
-
CSCF and S
-
CSCF but the result is a completely balanced load on all S
-
CSCFs of the
network.

So,

t
he cost of change for the network
is only some minor software changes.

On
the other hand,
the benefits one receives by the introduction o
f both changes are much
greater, and these are speed, reliability and higher utilization of the network.



23

Alexandros Fragkopoulos





ICTE 8




Chapter 5


Implementation


After having analyzed many different aspects of the IMS networks and prese
nted two papers
that actually propose improvements in two important entities of the IMS framework, the
project will go further in presenting an implementation inspired by these two pr
oposals
mentioned in

the two different use cases in

chapter 4.
In this ch
apter the implementation
will be shown and important aspects of it will be discussed. Results will also be given and
compared with different cases.


5.1 Description of the algorithm


For the needs of the implementation a
n algorithm was made to simulate

the procedure of a
user that gets connected to a P
-
CSCF and t
hen to the S
-
CSCF where a significant part

of the
intelligence of an IMS network is located. In order to study some differences between
different scenarios, some
parameters of the algorithm

were

changing so it could be possible
to create four different scenarios and then compare them.

These four scenarios are the following:

1.

Random search of
an available slot in a

P
-
CSCF

node

for a user and then
random search of a
n available slot in an
S
-
CSCF

node
.

2.

Search of a
n available slot in a

P
-
CSCF based on the minimum possible
delay of all the available P
-
CSCF nodes and then

random search of a
n

S
-
CSCF
.

3.

Search of a
n available slot in a

P
-
CSCF based on the minimum possible
delay of all the available P
-
CSCF nod
es
and then search of
an available
slot in an

S
-
CSCF based on the minimum possible delay of all the
available S
-
CSCF nodes.

4.

Random search of a
n available slot in a

P
-
CSCF for a user and then
search of
an available slot in an

S
-
CSCF based on the minimum pos
sible
delay of all the available S
-
CSCF nodes.


In order to
simulate

these scenarios and get a realistic sense in our results, some things
should be taken into consideration in making the algorithm.

So, before this project go any
further, a clarification should be given on what exactly the delay takes into consideration in
this work for both the user and the node side.

This delay is a
factor

that represents:

1.

For users: signal quality, processing time

or other factors that the user’s position
or device is responsible for the delay.

2.

For nodes: the node’s distance from the user, the load of the node or other
parameters

that affect the node’s processing time and acknowledgement of the
user’s request.

24

Alexandros Fragkopoulos





ICTE 8




So,
given

the

aforementioned

clarifi
cation
, the algorithm should have a list
of users that
request

a connection to IMS servi
ces. These users have a
delay from zero to one and it is
based on a uniform distribution, meaning that in the given interval the probabi
lity, of any
value is the same. T
he algorithm is

making use

also

of

a delay for the node
s of the P
-
CSCF
and the S
-
CSCF based also on a uniform distribution with a value range of zero to one.

In
order to give a more realistic sense to the who
le program, the

distributions for the delays of

the P
-
CSCF and the S
-
CSCF nodes are changing after every iterat
ion, meaning that the nodes
have

longer or s
horter distance from each user


since the users are moving


and that their
processing time is varying in time.

We
also assume that the S
-
CSCF’s capabilities are
adequate for the service that the user wishes to use.

The algorithm

function
s as introduced

in the following steps:

1.

The number of users is 500 and the number of available slots

is

200 in ten nodes

for
each entity

of twenty free slots each.

2.

The user requests a connection with a P
-
CSCF
.

3.

Search

f
or

a Proxy based on the node with

the minimum delay

or a random search
(it depends which scenario it is studied)

4.

If all the P
-
CSCF nodes are full then the pro
gram retries a few times in case

a node
had a free slot but it wasn’t found yet (that applies only to the random search
schemes)

and then exits.

5.

If a P
-
CSCF node is found for the user then the algorithm proceeds in finding a
n

S
-
CSCF.

6.

Search

for

a Server is

done based also on the node with

the minimum delay

or a
random search (it depends which scenario it is studied)

7.

If all the S
-
CSCF nodes are full then the program retries a few times in case
something was not done correctly and then exits but before exitin
g, it erases the
user from the P
-
CSCF connected list.

8.

When the maximum number of users that can be connected is reached the program
retries a few times in
case a node had a free slot but it wasn’t found yet (that applies
only to the random search schemes)
and then exits.

9.

As long as this procedure takes place
,

five plots
are showing in every iteration the
following measurements:

a.

The amount of users that are connected to a P
-
CSCF node and in which node
they are connected to (top left position)

b.

The total delay

of a user to get connected to a P
-
CSCF and to a
n

S
-
CSCF and
this user’s own delay (top middle position). Above this plot one can also see
the current total delay of each user
as he gets connected after each

iteration.

c.

The amount of users that are connecte
d to a S
-
CSCF node and in which node
they are connected to (top right position)

d.

A normal distribution which depicts the delay that most users have in order
to get connected until an S
-
CSCF (bottom left position).

Above this plot one
can also see the averag
e total delay after every iteration.

e.

How many users are connected to an S
-
CSCF and in which iteration this
happens (bottom right position). Above this plot one can also see the
current number of total connec
ted users after
each

iteration
.



25

Alexandros Fragkopoulos





ICTE 8




5.2
Figures and

Results


The algorithm produced

some plots that will be
illustrated

in the f
ollowing pages. Above
each plot

there will
be a description where one can

see
which the scenario is and the

measurements

that

were taken
.


First C
ase

In this scenario

(
Figure
3
)

the user is connected with a random selection of a P
-
CSCF and a
random
selection of a S
-
CSCF.


Fi gure
3



Pl ot s f rom Fi rst Case Scenari o


This algorithm needed
111.3922

seconds to finish its users’ registration.

Note that the ‘User
Delay: 0’ shows that there was no user registered in the last iteration of the program so the
variable remained without a value
.



26

Alexandros Fragkopoulos





ICTE 8




Second

C
ase

In this scenario

(
Figure
4
)

the user is connected with a selection based on the minimum
possible delay to a P
-
CSCF and a random selection of an S
-
CSCF.



Fi gure
4

-

Pl ot s f rom Second

Case Scenari o


T
his algorithm needed
75.7921

seconds to finish its users’ registration. Note that the ‘User
Delay: 0’ shows that there was no user registered in the last iteration of the program so the
var
iable remained without a value.


27

Alexandros Fragkopoulos





ICTE 8




Third

C
ase

In
this scenario

(
Figure
5
)

the user is connected with a selection based on the minimum
possible delay to a P
-
CSCF and a selection based on the minimum possible delay to
a S
-
CSCF.



Fi gure
5

-

Pl ot s f rom
Thi rd

Case Scenari o


This algorithm needed
70.5785

seconds to finish its users’ registration. Note that the ‘User
Delay: 0’ shows that there was no user registered in the last iteration of the program so the
variable remained without a value.


28

Alexandros Fragkopoulos





ICTE 8




Fourth Case

In this scenario

(
Figure
6
)

the user is connected with a

random

selection
of

a P
-
CSCF and a
selection based on the minimum possible delay to a S
-
CSCF.



Fi gure
6

-

Pl ot s f rom Fourt h Case Scenari o


This algorithm needed
110.5606

seconds to finish its users’ registration. Note that the ‘User
Delay: 0’ shows that there was no user registered in the last iteration of the program so the
var
iable remained without a valu
e.


5.3 Results’ Analysis


The following table

(
Table
1
)

shows some importa
nt measurements derived from the

scenarios

presented above
.

Tabl e
1



Acqui si t i on of dat a f rom t he si mul at i ons


First
Case

Second Case

Third Case

Fourth Case

Users Connected

195

183

193

200

Elapsed Time

111.3922

75.7921

70.5785

110.5606

Av
erage Connection T
ime

1.4659

1.0588

0.6169

1.1107


29

Alexandros Fragkopoulos





ICTE 8




In most of the cases (except the total number of connected users), the third case (selection
of node with minimum delay in both P
-
CSCF & S
-
CSCF entities) is better. The performance of
this approach of an IMS
network

is improved by applying the selection of

a node based on its
minimum delay in both P
-
CSCF and S
-
CSCF.

The following figure (
Figure
7
)

derived from the sum of the normalized values of the table
above and shows clearly which case achieves the best results.



Fi gure
7

-

Ranki ng of t he Best Case Scenari o


As it was stated above, the algorithm in the third case
decr
eased

the
delay

by 46.22
% from
the first case where every selection in each entity wa
s random. T
he only parameter that the
first case

had

exceeded

by

the third

was the number of users that assigned during the
process but the difference was only 1.01%

more users compared to the third one
. The third
case had
also
an exceptional average time of connection (until an S
-
CSCF) for each user

(
57.91
% lower delay from the first case)

since the algorithm was searching in both entities
the connection path with th
e lowest available delay factor.

Even the elapsed time for the
algorithm to finish directing all the users to the available slots
, was smaller than all

the other
cases. With respect to the first case the third case was faster by
36.64%. The overall
behavio
r of the customization in the third case
is the optimal one in respect to the other
three cases.



2.025

1.487

1.089

1.750

0
0.5
1
1.5
2
2.5
First Case
Second Case
Third Case
Fourth Case
Total Number for each case

30

Alexandros Fragkopoulos





ICTE 8




Chapter 6



Conclusions


Along this report many topics were discussed c
oncerning the IMS networks. Moreover
, two
papers were presented and a suggestion that

the combination of them could improve even
more the IMS framework, in terms of performance and reliability
, was given
. Finally, an
approach inspired from these two papers was introduced and some results derived, showing
that in a system where randomness i
s narrowed

down

and information is exchanged, the
system becomes more reliable and faster so the final user experience and the QoS are
increased.

The only disadvantage in order to exchange information that will make the
system more intelligent, flexible an
d more reliable, is that one has to add overhead in
messages.

In the next subchapter there will be a reference to the LTE

and LTE
-
A

networks
and what users should expect in the near future.


6
.1

Future
Vision in Communication Technologies


IMS is an
emerging architectural framework, as the 4G networks will push things for better
QoS

and QoE

and more demanding services but still there is way to go. Currently, circuit
switched networks are still in use


maybe phasing out but still used by a big part of

the
market.

LTE Networks

Since the LTE (Long Term Evolution) Networks will be the next big thing in the area of mobile
communications, it is only right to present some of the key aspects of them.

LTE networks were first started to be developed as an evolu
tion of 3G mobile system in
2004. Later, in the 8
th

release of 3GPP the form was standardized and in January 2008 it was
approved and finalized. Since then minor changes are made to this. Some of the
requirements for LTE are that for high mobility scenario
s, the speeds would vary around
100Mbps in DL and 50Mbps in UL with a 20MHz

bandwidth. One other requirement is that
the QoS should be enhanced for end
-
to
-
end services and also that there should be flexibility
concerning the spectrum, varying from 1.25 to
20MHz

[15]
.
The LTE use
the OFDM
(Orthogonal Frequency

Division

Mul
tiplex
)

modulation

for the physical layer and for the
transmission

MIMO (Multiple Input Multiple Output)

antennas system
, as it performs a lot
better than SISO
(Single

Input Single Output
).
Although for most people LTE is the
specification that complies with

the name of 4G, in fact this is not correct. With the
improvement of LTE


LTE advanced

, the connection is now correct. LTE advanced among
other things wil
l support higher speeds for scenarios

of low but also high mobility

[9]
.



31

Alexandros Fragkopoulos





ICTE 8




Bibliography

1.
Shyam Chakraborty, Tomas Frankkila, Janne Peisa, Per Synnergren.

IMS Multimedia
Telephony over Cellular Systems.
s.l.

:
John Wiley & Sons, Ltd., 2007.

2.
Challenges to the Next Generation Services in IP Multimedia Subsystem.
Kai
-
Di Chang,
Chi
-
Yuan Chen,Jiann
-
Liang Chen, Han
-
Chieh Chao.

2, s.l.

: Journal of Information
Processing Systems, 2010, Vol. 6. 1976
-
913X.

3. SIP Mess
ages.
Voip Think.
[Online] [Cited: April 26, 2012.]
http://www.en.voipforo.com/SIP/SIP_messages.php.

4.
Himanshu Pant, Andrew R. McGee, Uma Chandrashekhar, Steven H. Richman.

Optimal
Availability and Security for IMS
-
Based VoIP Networks.
Bell Labs Technica
l Journal.
2006,
Vols. 11(3), 211

223.

5.
Call and Messaging Performance Comparison Between IMS and SIP Networks.
Jari
Kellokoski, Erno Tukia, Eero Wallenius, Timo Hämäläien.

Jyväskylä

: IEEE, 2010.

6.
IMS interworking using IBCF .
R. Vargic, Krhla M., Sch
umann S., Kotuliak I.

s.l.

:
Convergence and Hybrid Information Technology, 2008. ICCIT '08. Third International
Conference, 2008, Vols. 2 (89
-
94).

7.
J. Peterson, H. Liu, J. Yu, B. Campbell.

IETF Documents.
[Online] June 2004. [Cited:
April 26, 2012.] htt
p://tools.ietf.org/html/rfc3824.

8.
Active location reporting for emergency call in UMTS IP multimedia subsystem.
Meng
-
Hsun Tsai, Yi
-
Bing Lin, Hsiao
-
Han Wang.

12 (5837
-

5843 ), s.l.

: Wireless
Communications, IEEE Transactions on, 2009, Vol. 8.

9.
Asif, S
aad Z.

Next Generation Mobile Communications Ecosystem
-

Technology
Management for Mobile Communications.
s.l.

: John Wiley and Sons, Ltd., 2011.
9780470747469.

10.
Performance Analysis of IP Multimedia Subsystem (IMS) Signaling in Heterogenous
Access Netw
orks.
Shyam S.Wagle, Nikolaos Doulamis, Minesh Ade, M. Ghazanfar Ullah.

2011

: IEEE.

11.
IP
-
Multimedia Subsystem (IMS) Performance Evaluation and Benchmarking.
Wagdy
A. Aziz, Salwa H. EL
-
Ramly, Magdy M. Ibrahim.

Irkutsk Listvyanka

: IEEE Region 8
SIBIRCON
-
2010, 2010.

12.
Project, 3rd Generation Partnership.

3GPP TS 32.454 V11.0.0 (2011
-
12) (11th
Release).
s.l.

: 3rd Generation Partnership Project, 2011
-
12.

13.
Notification based S
-
CSCF load Balancing in IMS Networks.
Isam Abdalla, S.
Venkatesan.

Dallas

: IEEE, 2011.

14.
Efficient P
-
CSCF Discovery through NASS
-
IMS.
Wu, Chen.

Shanghai

: Second
International Conference on Education Technology and Computer (ICETC),
2010.

15.
Seddik, Karim.

LTE and LTE
-
Advanced:An Introduction. Alexandria University & Nile
University

: s.n., 2011.

32

Alexandros Fragkopoulos





ICTE 8