A Generalized Model Transformation approach

to Link Design Models to Network Simulators

NS-3 Case Study

Iyas Alloush

1,2

, Yvon Kermarrec

1,2

and Siegfried Rouvrais

1,3

Telecom Bretagne, Institut Mines-Telecom, Universit

e Europ

eenne de Bretagne, Brest, France

UMR CNRS 6285 Lab-STICC, Brest, France

IRISA, Rennes, France

Keywords:

Model Driven Engineering, Simulators, NS-3, Telecom Service, Enterprise Architecture, Veriﬁcation, Model

Transformation, Code Generation, Tool Chains.

Abstract:

Telecom service creation (TSC) activity is one of the most important phases of a TS life cycle. There are

many efforts that were done to improve this activity recently. The early veriﬁcation of the TS from its design

models give an advantage to the service provider to improve the qualities and to detect design errors before

the implementation phase. Simulation makes it possible to predict the system behavior avoiding the cost

of real systems. Our objective in this paper is to present our methodology to link high abstract models of

telecom services to network simulators. Relying on Model Driven Engineering, we propose a generalization of

code generation methodology using an IMS meta-model and simulator-dedicated templates. In our approach,

the network simulator speciﬁcations are related to the transformation template only, while the underlying

platform speciﬁcations and standards are included in the meta-model. We illustrate our approach with a new

transformation to generate conﬁgurations for NS-3. We apply an example of a video conference service to

generate the simulation code.

1 INTRODUCTION

In the scope of the Telecom Service (TS) Veriﬁca-

tion (Combes and Renard, 1999), the service designer

needs to verify the service design if it satisﬁes the pre-

deﬁned funtional and non-functional (Chung et al.,

1999) requirements or not. This veriﬁcation activ-

ity is better to be performed as early as possible, so

to avoid the expensive consequences that may happen

when correcting the design errors and ﬂaws after the

software and hardware installation.

We aim to assist the different stakeholders that

are involved in the TS design (Combes and Renard,

1999), and more speciﬁcally, to help them detect the

design errors and ﬂaws, so to improve the qualities

(Chiprianov, 2012) such as the performance, cost, etc.

Network simulators provide valuable feedback in

the behavior of Telecom Services. Although, there

are simulators which have friendly user interfaces and

modeling environments, they still consume the de-

signer time due to the complexity of the design ar-

chitecture. The domain speciﬁcity of network sim-

ulators, the daunting coding, and long time conﬁgu-

ration, makes it difﬁcult to verify the TS through its

design models manually.

The difference between the technologies that are

used to design the telecom service architecture and

the ones that are used to model it in the simulator envi-

ronment raises a question: How to link directly and

automatically between these two technical spaces

(design and simulation)?

Observing the simulators’ output makes it possi-

ble to check design errors and quality violations. Our

objective in this paper is to provide the TS designer

with a tool that links directly the design model to the

simulator environment. Therefore, we can obtain spe-

ciﬁc measurements that will be needed in further veri-

ﬁcation activities according to the functional and non-

functional requirements.

Our recent work (Alloush et al., 2012) illustrated

our approach in using Model Driven Engineering

(MDE) to link the high abstract models to simula-

tors (e.g. OPNET). We rely on the core-network plat-

form IP Multimedia Subsystem (IMS) to perform the

video conference functions. IMS provides the ability

of deploying large number of Telecom Services that

337

Alloush I., Kermarrec Y. and Rouvrais S..

A Generalized Model Transformation approach to Link Design Models to Network Simulators - NS-3 Case Study.

DOI: 10.5220/0004407503370344

In Proceedings of the 3rd International Conference on Simulation and Modeling Methodologies, Technologies and Applications (SIMULTECH-2013),

pages 337-344

ISBN: 978-989-8565-69-3

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

are composed of different types of software applica-

tion thanks to its advanced structure that mediates be-

tween different telecom evolutions. We have obtained

manually the measurements that we relied on to check

the behavioral view of the TS design. The design ar-

chitecture of the TS was automatically generated by

our tool relying on Model Driven Engineering (MDE)

and Eclipse Modeling Framework (EMF) in Eclipse.

The measurement settings were manually conﬁgured,

and the measurement analysis were done manually to

check the behavioral view of the TS design. Addi-

tionally, relaying on one simulator to test the design

is not enough, as simulators differ from each other in

their measurement capabilities and certiﬁcate levels.

In this paper, we generalize and develop our ap-

proach to be used with domain speciﬁc (network) sim-

ulators, and we include the automatic generation of

the measurement conﬁguration.

Our ﬁrst contribution in this paper is in extend-

ing the Meta-Model (MM) that is presented in (Al-

loush et al., 2012; Chiprianov et al., 2011) to include

new elements that improve the code generation abil-

ities to adapt with different network simulators. We

add new measurement and tool (simulator) entities to

that MM to enable the automatic generation of the

measurement conﬁguration. We argue, as our second

contribution, that by ﬁxing the MM which represents

the core-network platform (e.g. IMS), we can gener-

ate the code that is needed to conﬁgure different net-

work simulators by changing the transformation tem-

plate using Eclipse environment. We illustrate our ap-

proach with an example of conﬁguration generation to

the NS-3 network simulator, where as far as we know,

there is no similar work done using NS-3 yet. This

is added to the previously-argued case study in (Al-

loush et al., 2012) that is speciﬁc for another network

simulator (OPNET) to illustrate the generality of our

approach.

In section 2,we present brieﬂy the related work,

highlighting the points of interest to our objective. In

section 3, we will provide an explanation about model

transformations that we rely on to generate the code.

In section 4, we highlight the Enterprise Architecture

(EA) with its architecture, and show the beneﬁts from

applying its structure to our approach. In Section 5,

we present a short explanation about network simula-

tors, and show the features of the different simulators

that we used in our work. Section 6 will be dedicated

for our contributions, where we present our general-

ized approach to obtain wide range of measurements

for the veriﬁcation process. Then we present the new

extensions of the Meta-Model, and we give a brief

view to the NS-3 code generation method. In Section

7, we discuss an example of a session creation for a

video conference TS, presenting the new model in-

stances that are needed to generate the measurement

conﬁguration. Then we conclude in section 8 and dis-

cuss our future work.

2 RELATED WORK

This paper is in the scope of TS creation activity. As

our work is a continuation to (Chiprianov, 2012) and

according to the requirements of the early test activity,

we will highlight some of the others work from this

domain, taking into consideration the following crite-

ria: (1) Involving the different actors during the ser-

vice creation activity; (2) the Platform Speciﬁc Mod-

els (PSMs) level of details with structural elements

and if it is enough to start simulation activity or not;

(3) the ability to generate code directly from Plat-

form Independent Models (PIMs); (4) The possible

execution or simulation environments; (5) Usage of

Domain-Speciﬁc tools; (6) The ability to collect mea-

surements and to conﬁgure them automatically.

In the following, we present some of the work that

is related to early veriﬁcation activity before the im-

plementation:

• In (Achilleos et al., 2008), the authors proposed

an approach for service creation that introduces

a service validation straight after the service de-

sign phase and before the implementation. Their

methodology integrates the Model Driven Archi-

tecture (MDA) and Petri Nets to provide design,

validation and code generation process.They use

Petri Nets to validate the service behavior and en-

sure its correctness. Their approach involves the

End-User in the ﬁnal validation process, where

the validation software is obtained by the same

generative approach and MDA. Their methodol-

ogy provides behavioral details only, and presents

no relationships between different nodes of an ex-

ecution network. This prevents the tester from

being able to generate a network simulation sce-

nario. They generate Java code for their exam-

ple directly from the PIM interface model. The

presented Meta-Model of Service Oriented Petri

Net shows no possibility to analyze the behavior

of the service using any other method else than

Petri Nets;

• In (Hartman et al., 2007), the paper describes a

model-based approach to create telecom services.

They rely on the code generation to implement

the service relying on IP Multimedia Subsystem

(IMS) infrastructure. The stakeholder who is in-

volved in their approach is the service provider.

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

338

They start from the high abstract models as PIM

to deﬁne the service functions using state ma-

chine UML2 models. Using Eclipse software,

they could close the abstraction gap between the

design and the execution environment, by hid-

ing the low level protocol and architecture details.

The code generation process produces Java pack-

ages (e.g. SIP servlet) that can run on a SIP en-

abled Application Server (e.g. IBM WAS version

6.1). Simulation activity is not included as there

is no hardware entities descriptions in the PSM;

• The work presented in (Adamopoulos et al., 2002)

addresses the same scope of service creation ac-

tivity. They propose a methodology that takes into

consideration the customer and service provider

requirements. The validation activity is per-

formed after the implementation phase, and the

service speciﬁcation is represented by use-case

models as PIMs. They apply code generation to

produce Java and C++ codes directly from these

PIMs. No simulators were used;

• The paper (Touraille et al., 2011) presents work

that is very close to the formalism which we use.

The authors apply MDE and rely on DEVS (Dis-

crete EVent Systems Speciﬁcation) formalism to

smooth the modeling and simulation cycle and

facilitate the coupling of models using heteroge-

neous formalisms. They provide the SimStudio

software which aims at providing modeling and

simulation tool chain. This assists the developer

in all of the activities from the design to the result

analysis. They produced an advanced tool chain

that contains 4 axis: modeling, analysis, visual-

ization, and Management. The SimStudio relies

on DEVS that contains a simulation kernel which

offers the possibility of high abstract level of mod-

eling. Their approach and ours depend on meta-

models that contains different views to the differ-

ent activities of the tool chain (modeling, simula-

tion). Although they have analysis tools that we

didn’t include in our tool chain, their tool chain

doesn’t have support for structural elements rep-

resentation while limited to the behavioral ones.

Their approach doesn’t show domain speciﬁcity,

where we rely on IP Multimedia Subsystem (IMS)

platform. The different views of the actors of an

enterprise are not considered as it shows in their

approach.

In this paper, we will illustrate how our work satisﬁes

all of the previously mentioned criteria.

3 MODEL TRANSFORMATIONS

Model transformation is one of the key concepts

in MDE, and in our approach. There are different

types of model transformations (van Amstel, 2010;

Mens and Van Gorp, 2006) and they are categorized

due to different views. There are different types

of model transformations: model-to-model, text-to-

model, model-to-text (which we use), and text-to-

text. From the meta-model or grammar change view

point, there are two types of model transformations:

endogenous model transformation where the source

and target models conform to the same meta-model,

and the exogenous model transformation where the

source and target models conform to different meta-

models. Our model transformation is an exogenous

model transformation, as the target is a conﬁguration

script for a simulator with different syntax rules and

constraints. The difference in syntax and semantics

between the two technical spaces (design and Simu-

lation) creates a challenge to link between them. In

our approach, the code generation rules are used to

equilibrate this difference (Chiprianov et al., 2010).

XPAND is a model-to-text transformation lan-

guage that makes it possible to iterate over input

model instances to generate text ﬁles. It has the ca-

pability of collecting the data of the design model and

insert it in the form of text embedded in the different

statements of the code that is used to conﬁgure the

target tool. This enables us to generate Java packages

for high abstract models (PIMs) (Chiprianov, 2012),

and conﬁguration scripts for simulators for the plat-

form speciﬁc models (PSM). We tend to make the

XPAND templates as far as possible from the hard

coding, this implies to deﬁne the needed entities in

the Meta-Model and so in the input model.

4 ENTERPRISE ARCHITECTURE

(EA) AND ArchiMate

An Enterprise Architecture (EA) ”is an instrument to

articulate an enterprise’s future direction, while also

serving as a coordination and steering mechanism

toward the actual transformation of the enterprise”

(Greefhorst and Proper, 2011). In our approach, we

apply the EA standard into the TS architecture (Si-

monin et al., 2007). EA framework provides a tele-

com enterprise a way to decompose a complex archi-

tecture of a TS into aspects and layers. The aspects

dimension provides an aspect conceptualization of an

enterprise. While the layer dimension separates the

TS architecture into 3 different layers according to the

level of details.

AGeneralizedModelTransformationapproachtoLinkDesignModelstoNetworkSimulators-NS-3CaseStudy

339

Figure 1: Enterprise Architecture and Code Generation in

our approach.

In our approach, we apply ArchiMate which is

an EA modeling language, as ArchiMate seems to be

suitable for IT domain modeling (Chiprianov, 2012).

Figure (Fig.1) presents the two dimensions of Archi-

Mate, and illustrates the separation in abstraction be-

tween the different layers from our view point. Re-

garding to the criteria in section 2, EA and ArchiMate

provides a solution for the simulation needs, as it: (1)

separates between the aspects of behavioral and struc-

tural entities; (2) provides designers with 3 layers that

are different in the level of details, which makes it

possible to specify the models of the technology layer

as PSMs; (3) involves the different stakeholders in the

design activity. This answers to the criteria 1, 2 and

3 that are mentioned in section 2. The dedication of

layers due to the domains makes it possible to involve

the different stakeholders of a service creation activ-

ity that is mentioned in (Combes and Renard, 1999).

We consider the service provider, designer, and ser-

vice developer stakeholders in our work, as an answer

to the ﬁrst criteria in section 2.

5 NETWORK SIMULATORS

System modeling makes it possible to modify the de-

sign parameters and to test the system and analyze it

more than once, without adding experiment cost.

5.1 Network Modeling Approaches

To model a system for testing purpose, some simpli-

fying assumptions are often required. Relying on too

many assumptions for simpliﬁcation purpose, may

lead to an inaccurate representation of the system. In

the scope of computer networks, there are two mod-

eling approaches (Issariyakul and Hossain, 2009):

1. Analytical Approach. This approach relies on

mathematical description for the system with the

help of mathematical tools such as queuing and

probability theories, then applying that descrip-

tion using the numerical methods. Such an ap-

proach may not give accurate representation of the

Figure 2: Our approach to verify Telecom Services after the

design activity relying on tool chains.

real system if many simplifying assumptions were

made;

2. Simulation Approach. This approach requires

fewer simplifying assumptions (less abstract level

in the model) than the analytical approach. One

can set every needed speciﬁcation in its place to

describe the actual system in the best way.

Both approaches may leave out some details, since

having a lot of details may result in large computa-

tional efforts and unmanageable execution. In our ap-

proach, at the design activity, we need to model archi-

tectures, with both structural and behavioral entities

in high level of abstraction. Thus, we have chosen the

simulation approach for modeling the TS design, as

we prefer to have a better description of the TS de-

sign and to execute its behaviors in an accurate way.

5.2 NS-3 Simulator

We select NS-3 as it: (1) supports tracing functions, to

collect the measurements and export them into ﬁles;

(2) accepts conﬁguration using Python and C++ for

front-end (e.g. scripting, visualization); (3) supports

different types of easy-conﬁgured applications (e.g.

Socket using TCP or UDP). NS-3 is a proper simula-

tor that ﬁts the nature of TSs, as it accepts (in addition

to the previous features) plain text input ﬁles, and gen-

erates measurements that can be analyzed to improve

the qualities of the TS.

6 CONTRIBUTIONS

6.1 Methodology

Our general approach relies on tool chains to achieve

a complete loop 2 between the design and early veri-

ﬁcation activities in Telecom Service Creation (TSC)

activity (Combes and Renard, 1999).

The veriﬁcation activity may require different

simulators according to the requirements nature and

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

340

variety, where the TS is going to be veriﬁed with re-

spect to these requirements. Therefore, we aim to pro-

vide the designer with a methodology that is able to

use wide range of simulators. In this paper, our con-

tributions have one broad aim (Fig.3) : to link the high

abstract design models from the TS domain to the

network simulators so to obtain measurements from

the modeled system and analyze in the next activities.

Therefore, we highlight two major contributions that

helps achieving this target:

• Extending the technology layer meta-model that

is mentioned in (Alloush et al., 2012) to include

new adequate entities that are relevant to the sim-

ulation activity and of general usage for different

types of network simulators;

• Proposing an approach based on MDE to link be-

tween design models and simulators using a gen-

eralized method, we ﬁx the Meta-Model (syntax)

and change the model transformation template

only to generate the code for the different simu-

lators. This is coherent with the advanced core-

network platforms (e.g. IP Multimedia Subsys-

tem IMS (Camarillo and Garc

ıa-Mart

ın, 2008)),

as changing the infrastructure for every new IT

service is not a practical or economic solution.

Normally, once the infrastructure is installed and

implemented, it is better for the network provider

to invest in it for long period.

6.2 Meta-model Extension

We present the technology layer Meta-Model exten-

sion that is needed to generalize our approach to link

between the high abstract models and the simulators.

The extension is tool (simulator) oriented, and in-

cludes the following actions:

1. Extending the NodeInterface entity (Fig.4) which

is deﬁned using ArchiMate language (The Open

Group, 2009) and of the type InfrastructureInter-

face. We have deﬁned more protocols that corre-

Figure 3: An MDE approach to link high abstract models to

simulators including measurement conﬁguration.

<<InfrastructuralInterface>>

NodeInterface

L2 Protocol

+name: String

L3 Protocol

+name: String

L4 Protocol

+name: String

L7 Protocol

+Name: String

Protocol

Network

+Name: String

+NetAddress: String

+NetMask: String

+id: String

Figure 4: NodeInterface extension to the IMS meta-model

presented in (Alloush et al., 2012).

spond to the OSI (Open Systems Interconnection)

system. This makes it possible to map the design

with different types of network simulators of dif-

ferent granularity levels;

2. We deﬁne the Measurement model (Fig.5) of type

TechnologyLayerElement from ArchiMate lan-

guage (The Open Group, 2009) that contains:

measurement function, measure, and the probe.

The probe (Chang, 1999) is the entity that is re-

lated to the simulator part, while the measure is

the register that saves the ﬁnal result of the data

manipulation after the call of the measurement

function. To enable the manipulation of the col-

lected data by different probes, we deﬁne mathe-

matical operators such as (Sum, Subtract, Multi-

ply, and Divide, etc). These operators are used

in the measurement function to enable the cus-

tomization of measurements by the domain ex-

perts (Fig.3). These measurements are linked to

the design entities (behavioral and/or structural)

through association relations (The Open Group,

2009). The reason behind this link is that there

are some measurements related to hardware ele-

ments (e.g. CPU utilization is related to nodes)

while (Session Start/End Probe is related to a be-

havior of exchanging control messages between

nodes). This contribution answers to the criteria 6

in section 2;

3. We deﬁne the ”associated with” relationship from

ArchiMate language (The Open Group, 2009) to

associate structural and behavioral entities of an

IMS platform with the corresponding measure as

a result of the measurement function that is pro-

cessed in the simulator.

The mentioned IMS Meta-Model and its extensions

are prototyped and implemented in an ecore ﬁle

(Meta-Model) using Eclipse EMF tool.

AGeneralizedModelTransformationapproachtoLinkDesignModelstoNetworkSimulators-NS-3CaseStudy

341

Measurement Function

Measure

Access

InfrastructureInterface(Archimate)

Associated with

0..*

Probe

+type: string

Measurement Model

+Name: String

1..*

Object

<<Tool>>

Simulator

+Measurements: List

+isCustomizedApplication: Boolean

+Probes: List

+Certificate Level: string

<<InfraStructure Service (Archimate)>>

TechnologyFunction

Associated with

0..*

1..*

Tool

Artifact

Figure 5: Measurement view from the tool speciﬁcation

Meta-Model linked with IMS MM presented in (Alloush

et al., 2012).

6.3 Using Multiple Network Simulators

The extended meta-model presented in subsection 6.2

have the same speciﬁcations that were used in the pre-

vious XPAND template that is used with OPNET tool

in (Alloush et al., 2012), and it works now also with

the NS-3 simulator template, where we use the same

model transformation language (XPAND) in section

3. OPNET and NS-3 are widely used in the domain

of network simulation, and the extended MM pro-

vides enough abstract syntax to implement and run

the model transformation. Additionally, the extension

depends on telecom standards and basics that every

network simulator should respect. This enables us

to answer the question of reusing our approach with

other network simulator.

Figure 6: Part of the XPAND template presents the itera-

tions to decompose and map the Service Design to NS-3

model.

Figure 7: Mapping elements between TS design and NS-3

technical spaces (implemented in XPAND).

6.3.1 Mapping TS Design Model to NS-3 Model

We will present the main conceptual modeling con-

cepts in NS-3 and its hierarchy, and how we mapped

it (Fig.7) to the IMS models that we have designed.

To build up the topology of the network and the ser-

vice structural design, we use the structural view from

the model of the TS instance. This model contains

the structural entities of the TS service, one of these

entities is the CommunicationPath that represent the

point-to-point connection between to NodeInterfaces.

Thus, iterating for every CommunicationPath (Fig.6)

makes it possible to identify the parent NodeInter-

faces, and then identifying the nodes that are joined

together using this CommunicationPath (Fig.10).

7 VIDEO CONFERENCE

CREATION EXAMPLE USING

IMS CASE STUDY

In this section, we illustrate our approach with an

example of a TS (Video Conference) relying on

IMS network platform (Camarillo and Garc

ıa-Mart

ın,

2008). We are interested in the control plane (sig-

naling) messages that are exchanged between the dif-

ferent nodes to perform the creation of the confer-

ence, and we focus on the application layer messages.

This is an important point for the designer. The Ses-

sion Initiation Protocol (SIP) (Camarillo and Garc

ıa-

Mart

ın, 2008) is the main application protocol in IMS.

SIP messages are text plain messages, the fact that

makes it easier to read and analyze them.

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

342

7.1 TS Design Model Instance

(Static View)

The design model of the TS (right part of Fig.10)

contains high abstract level of details such as Nodes,

application protocols, messages, sequence of func-

tions that call each other, etc. We present the struc-

tural view of the design and from the technology

layer speciﬁcally. The CommunicationPath entities

are clear in this example, we use them as a starting

point to identify the structure of the simulation sce-

nario as mentioned in 6.3.1. Relying on the types de-

ﬁned in the MM and the relations from ArchiMate,

we can investigate the topology of the scenario using

the iterative functions in XPAND (Fig. 6).

7.2 Examples of Measurements

and NS-3 Code Sample

Measurement instances should conform to the meta-

model (Fig. 5). For an IP network, we present 2

examples of measurement models that will be imple-

mented in NS-3 code automatically.

• DevQueueDrop: This is an example for a cus-

tomized measurement. This event is triggered

when a packet is dropped on the device level in

NS-3. We want to implement the trace of Queue

Packet Drop by deﬁning a counter that increments

by 1 every time the probe is triggered (Fig.8).

This mathematical operation is implemented in

the measurement function, where the inputs are:

the probe value itself and the constant 1;

• CongestionWindow: This measurement exists in

the NS-3 simulator, and their is no need to imple-

ment a measurement function for such a built-in

measurement (Fig.9).

DevQueueDrop Measure

Measurement Model

TraceDevQueueDrop, "+"

Access

DevQueueDrop, type=txQueue/DropDevQueueDrop

Access

Probe

Measurement FunctionMeasure

All(*): CommunicationPath

Asscociated with

From the design model

Constant

Figure 8: A model instance for DevQueueDrop Measure-

ment.

All(*),type=$ns3::TcpL4Protocol/SocketLis: L4Protocol

Congestion Window measurement model

Congestion Window: Measure

CongestionWindow: Probe

Associated with

Access

Probe

Figure 9: A model instance for CongestionWindow Mea-

surement.

Unlike OPNET, the conﬁguration method in NS-3

(NS3, 2012) makes it possible to use different types of

variables to be used inside the code for many purposes

such as calculations.

(Fig.10) presents the correspondence between the

generated code and the structural entities. Using (waf)

command we veriﬁed the result code with no errors in

the C++ rules neither in the NS-3 ones.

8 CONCLUSIONS AND FUTURE

WORK

In this paper, we have presented our approach to con-

nect the design models from the speciﬁcation activ-

ity to the network simulators directly. A Meta-Model

extension was proposed to generalize the linking be-

tween the modeling and network simulation techni-

cal spaces relying on the reusability of the extended

MM. We rely on the MDE approach to keep the ab-

straction level high at the design phase and to hide

the complexity from the designer. To verify the tele-

com service design early at the speciﬁcation phase,

our approach joins the two different technical spaces

(network simulation and TS design) together, attach-

ing the measurement entities that exists in the net-

work domain directly to the high abstract modeling

tools. Thus, we put the veriﬁcation utilities between

the TS designer hands, taking the tool speciﬁcity and

the daunting coding away from his responsibilities.

We have illustrated our approach with some examples

from the generated code and the corresponding input

models.

Our approach separates between the roles of the

technical spaces mapping, where the meta-model is

reusable with different types of network simulators

and the xpand template is related to the simulator

speciﬁcally. The usage of multiple simulators enable

the veriﬁer from testing the TS using wide range of

measurements. On the other hand, our method in gen-

erating the customized measurements takes into con-

sideration only two inputs for the measurement func-

tion element. Additionally, the implementation of the

XPAND template needs experience in the target sim-

ulator and modeling technical spaces besides to the

IMS platform ”execution platform”. These limita-

tions should be taken into consideration by the tool

vendor when adding new tools.

In the future, on the one hand, we intend to de-

ﬁne the method of tool selection. On the other hand

we will integrate between these tools so the data ﬂows

from one tool to another, this makes the tools comple-

ment each other in our tool chain.

AGeneralizedModelTransformationapproachtoLinkDesignModelstoNetworkSimulators-NS-3CaseStudy

343

Figure 10: C++ code sample as input of NS-3 mapped to the design model.

REFERENCES

(2012). NS-3 Manual-Release ns-3-dev, www.nsnam.org/

docs/manual/ns-3-manual.pdf, dev edition.

Achilleos, A., Yang, K., Georgalas, N., and Azmoodech,

M. (2008). Pervasive service creation using a model

driven petri net based approach. In Wireless Commu-

nications and Mobile Computing Conference, 2008.

IWCMC ’08. International, pages 309 –314.

Adamopoulos, D., Pavlou, G., and Papandreou, C. (2002).

Advanced service creation using distributed ob-

ject technology. Communications Magazine, IEEE,

40(3):146 –154.

Alloush, I., Chiprianov, V., Kermarrec, Y., and Rouvrais,

S. (2012). Linking telecom service high-level abstract

models to simulators based on model transformations:

The ims case study. In Information and Communi-

cation Technologies, volume 7479, pages 100–111.

Springer Verlag.

Camarillo, G. and Garc

ıa-Mart

ın, M. A. (2008). The 3G

IP Multimedia Subsystem (IMS) Merging the Internet

and the Cellular Worlds. Third edition.

Chang, X. (1999). Network simulations with opnet. In Sim-

ulation Conference Proceedings, 1999 Winter, vol-

ume 1, pages 307 –314 vol.1.

Chiprianov, V. (2012). Collaborative Construction of

Telecommunications Services. An Enterprise Archi-

tecture and Model Driven Engineering Method. PhD

thesis, Telecom Bretagne, France.

Chiprianov, V., Alloush, I., Kermarrec, Y., and Rouvrais,

S. (2011). Telecommunications service creation: To-

wards extensions for enterprise architecture model-

ing languages. In 6th Intl. Conf. on Software and

Data Technologies (ICSOFT), volume 1, pages 23–29,

Seville, Spain.

Chiprianov, V., Kermarrec, Y., and Rouvrais, S. (2010).

Meta-tools for Software Language Engineering: A

Flexible Collaborative Modeling Language for Efﬁ-

cient Telecommunications Service Design. In Flex-

iTools2010 ICSE Workshop on Flexible Modeling

Tools.

Chung, L., Nixon, B. A., Yu, E., and Mylopoulos, J. (1999).

Non-Functional Requirements in Software Engineer-

ing, volume I. KLUWER ACADEMIC Publishers.

Combes, P. and Renard, B. (1999). Service validation. Com-

puter Networks, 31(17):1817 – 1834.

Greefhorst, D. and Proper, E. (2011). Architecture Princi-

ples, volume 4 of The Enterprise Engineering series.

Springer.

Hartman, A., Keren, M., Kremer-Davidson, S., and Pikus,

D. (2007). Model-based design and generation of tele-

com services.

Issariyakul, T. and Hossain, E. (2009). Introduction to Net-

work Simulator NS2. Springer.

Mens, T. and Van Gorp, P. (2006). A taxonomy of model

transformation. Electron. Notes Theor. Comput. Sci.,

152:125–142.

Simonin, J., Le Traon, Y., and Jezequel, J. M. (2007). An

enterprise architecture alignment measure for telecom

service development. 11th Ieee International Enter-

prise Distributed Object Computing Conference, Pro-

ceedings, pages 476–483.

The Open Group (2009). ArchiMate 1.0 Speciﬁcation.

Touraille, L., Traor

e, M. K., and Hill, D. R. C. (2011).

A model-driven software environment for modeling,

simulation and analysis of complex systems. In Pro-

ceedings of the 2011 Symposium on Theory of Mod-

eling & Simulation: DEVS Integrative M&S Sympo-

sium, pages 229–237, San Diego, CA, USA.

van Amstel, M. (2010). The right tool for the right job:

Assessing model transformation quality. In Com-

puter Software and Applications Conference Work-

shops (COMPSACW), 2010 IEEE 34th Annual, pages

69 –74.

SIMULTECH2013-3rdInternationalConferenceonSimulationandModelingMethodologies,Technologiesand

Applications

344