AN ARCHITECTURAL FRAMEWORK FOR DELIVERING SIP-AS MULTIMEDIA SERVICES BASED ON JADE/OSGI TECHNOLOGY

International Journal of Next-Generation Networks (IJNGN) Vol.6, No.3, September 2014
AN ARCHITECTURAL FRAMEWORK FOR DELIVERING
SIP-AS MULTIMEDIA SERVICES BASED ON JADE/OSGI
TECHNOLOGY
Renato B. Cabelino Ribeiro1, Magnos Martinello2, Celso Alberto Saibel Santos3
and Rosane Bodart Soares4
1,4Departamento de Engenharia Elétrica, IFES, Vitória, Brasil,
2,3Departamento de Informática, UFES, Vitória, Brasil
ABSTRACT
This paper proposes a new scalable service-oriented architecture based on Open Service Gateway
Initiative (OSGI) technology. A key part of this architecture is its SIP application as a service (SIP-AS). It
relies on IMS core network supported by multi agents components implemented using Java Agent
DEvelopment (JADE) platform. As a proof of concept, a real testbed/prototype has been developed to
validate our approach. The validation process consisted of two phases: (i) configuration of the JADE/OSGi
SIP-AS architecture to provide a televoting service and (ii) characterization and analysis of jitter, packet
loss, load capacity and CPU utilization of the implemented architecture. Results demonstrate that this
televoting service scales up and out enabling the elasticity of the architecture on the processing of
concurrent calls and dynamic load balancing.
KEYWORDS
JADE, OSGi, IMS, multimedia services, televoting
1. INTRODUCTION
SIP (Session Initiation Protocol) has emerged as the signaling protocol designed for supporting IP
telephony and unified communications. The standard protocol is supported in a wide variety of IP
communications products including IP PBXs, servers, videoconferencing systems as well as desk
phones. Typically service providers have replaced their legacy telephony infrastructure by SIP-based
solutions and converging voice, video, and traffic onto a common IP backbone, thus they
can reduce costs and simplify operations.
SIP trunking services are offered as a cost effective alternatives to conventional PRI (Primary
Rate Interface) circuits for PSTN connectivity. By switching to a SIP trunking service, IT
organizations can eliminate TDM (time-division multiplexing) gateways, reduce monthly service
fees and improve service agility i.e. SIP trunks can be installed and reprovisioned more quickly
and easily than conventional PRI circuits.
The scalability of the architecture for providing SIP as a service is a huge challenge. In a
centralized topology, SIP is installed in central data centers. External calls are received from the
corporate WAN, aggregated, and handed off to the SIP service provider. Inter office calls are
carried over the corporate WAN. However, the rate of calls can vary significantly affecting the
quality of the provided service. Thus, network planners should evaluate carefully the service
architecture in order to determine which model best addresses their organization’s specific
DOI : 10.5121/ijngn.2014.6301 1


requirements. This paper introduces a service-oriented architecture combined with multi-agents
systems as a hosting platform for telecommunication supplementary services on IP Multimedia
Subsystem (IMS).
The JADE platform, integrated with the OSGi framework, is the proposed agent-based
development environment [1][2]. This approach allows for a more flexible and dynamic form of
service provisioning over the IMS architecture, allowing services to be negotiated on demand
according to the current environment requirements (services rules, QoS requirements, interaction
parameters, etc.).
This article presents the implementation of a SIP Application Server (SIP-AS) on OSGi service-oriented
architecture, integrated with JADE framework for the creation and provision of
multimedia services on IMS. In section 2, the technology is described. The related works are
presented in Section 3. Section 4 describes the design and implementation of the televoting
service, and the result analysis. Conclusions are described in section 5.
2
2. TECHNOLOGY BASE
2.1. IP Multimedia Subsystem (IMS)
IMS is an overlay service-provisioning platform, which allows telecommunication operators to
utilize Internet technologies to build an open IP-based service infrastructure that enables easy
deployment of new multimedia services [3][4].
Moreover, IMS uses the SIP protocol for session control and the RTP/RTSP protocol for media
streaming. Network requirements such as roaming, scalability, security and reliability have been
defined by 3GPP. The SIP protocol has been standardized by IETF - Internet Engineering Task
Force, but not its related architectures.
The layered approach proposed by IMS increases the importance of the application layer as
services are designed to work independently of the access network. IMS is designed to bridge the
gap between them [4]. It offers more flexibility for telecom operators to manage different services
with distinct requisites (e.g.: bandwidth, latency, jitter, etc.). The major IMS elements related to
service architecture are the following [3][4][5]:
• S-CSCFs (session management and routing family): the serving call state control functions
facilitate the correct interaction between the application servers, media servers, and the Home
Subscriber Service (HSS).
• HSS (database): the home subscriber server is the main data storage for all home subscriber
and service-related data of the IMS.
• ASs (service functions): the application services are entities that provide value-added
multimedia services in IMS, such as presence and Push to Talk Over Cellular.


3
Figure 1 presents these major elements of the IMS architecture.
Figure . Minimal IMS based architecture [4].
2.2. JADE/OSGi integration
JADE is a middleware for the development and execution of peer-to-peer applications based on
the agent paradigm that can easily work and interoperate on traditional or wireless network
environments. JADE internal architecture is currently the only architecture entirely compliant
with FIPA standards [6]. According to [7] and [8], the JADE platform can offer the following:
• Graphical interface which allows monitoring, debugging and logging;
• Components which can be distributed over the network;
• Mobility and cloning of agents as well as multi-tasking scheduling;
• Lifecycle management, name and yellow pages services, point-to-point message transport
service, speech-act message structure and ontology service;
• Interoperability with other platforms that offer support to FIPA standards.
Along with the agent-based environment, the OSGi framework allows for runtime installing,
updating and uninstalling of JAVA modules. The latest specification of the OSGi framework
defines the way collaboration among services offered by modules inside a single Java Virtual
machine (JVM) occurs.
The service platform offers a standard and open architecture to service providers, developers,
software developers, telecommunications operators and equipment manufacturers so that they can
develop and manage services in a coordinated and efficient way [9].
The development of applications using OSGi can be accomplished through the combination of
collaborative, reusable modules associated with descriptive information on their metafiles which
include service-related input that must be instantiated/imported to achieve a consistent execution
of the modules [9]. Also, the services provided by OSGi implement a JAVA interface for


registering on local service registries. Through this centralized control model, the modules (or
bundles) can verify their service dependencies.
Furthermore, the OSGi services platform offers developers the means to maximize the use of
platform independent resources and dynamic updating of JAVA modules, allowing development
of services for devices with limited computing resources, widely used in corporate environments.
New services registration as well as research and maintenance of pre-existing services (including
their uninstallation from the system), services status notifications and follow-ups on bundles
lifecycle can be carried out in a simple and efficient way.
4
3. RELATED WORKS
This section analyzes exclusively SIP approaches that have correlation with customized
telecommunications services (for example, additional services, toll free phone services,
Televoting, local number portability) or with the framework presented in this work. Oliveira et al
[10] propose two approaches for implementing number portability service in IMS networks,
tested on an AS (Application Server) according to the standards of the General Regulations of
Portability. In the first approach, the AS performs the function of local number portability without
call states control, routing every call originated by IMS to AS, or those based on a numeric phone
context through an Initial Filter Criteria (IFC). In the second approach, the number portability
service acts as a back-to-back user agent (B2BUA), i.e. in a leg termination call aimed to a ported
user, configures an IFC to conduct the call routing for an AS, which acts on behalf of the user
ported and initiates a new call to the correct destination. For the implementation of AS, SIP
Servlet technology was created by the authors.
Munadi et al [11] propose the design and implementation of VoIP services with OpenIMS and
ASTERISK, interconnected by an ENUM server which develops numerical mapping function
between the two servers. The authors observed the proposed environment according to: (a)
performance measures for each server; (b) the Post Dial Delay (PDD) and (c) of the same CPU
consumption. The values measured and analyzed in (a) identify the service time consumption on
the part of the SIP signaling system. In (b), three scenarios were tested where the Traffic
Analyzer WIRESHARK was used in order to capture and analyze traffic from the User Agent
Caller from your application until its acceptance by the counterparty in the call, which allowed
the analysis of the PDD in each test performed. For (c), were used the TOP utility from the
operating system itself in order to obtain the maximum CPU value throughout the experiments.
Li et al [12] implement two IMS services – a chat room (SIP-IM) and Presence services, in a SIP-AS.
SIP-AS architecture used is based on Mobicents SIP Servlet component (MSS). In addition,
we used the OpenIMS Core to IMS Core Network implementation. The authors have developed a
use case diagram and class diagram for the services analyzed. For the test scenarios, XML
templates were made to present the requirements and the design of both services.
Franks et al [13] describes the experiences and processes of the performance engineering of SIP
applications in Java environment using Java 2 Enterprise Edition (J2EE) as application server.
Two scenarios were evaluated: standalone and clustered environment. To build a performance
model of each environment, network traces were used. The predicted performance model was
calibrated with the data obtained from packet traces analysis to match with real system test data.
In both cases, the capacity predicted by the model was close to the capacity measured from
prototype system. Moreover, the performance model could help identify the bottlenecks in
performance of the system and was presented as a novel way of modeling Ethernet by using two-phase
fixed-rate server.


Femminella et al [14] describes a proposal and analysis of original usage of virtualization and
parallelization techniques in order to better exploit the computing capabilities of servers hosting
Java implementations of SIP application servers (ASs). The authors used a hypervisor to host SIP
ASs virtual machines (VMs) and an open source JSLEE AS (Mobicents) running a SIP-based
VoIP service, performing several database queries during an call lifetime. The main results shows
that the standard deployment of Java-based ASs do not provide the best performance, which is
obtained when the server runs in multiple systems in parallel, each one with few CPU cores
(running a single AS instance inside VMs, and deploying VMs with 32bit OS and 22/3 CPU
cores). With these scenarios the authors were able to evaluate a performance improvement of
about 64% in terms of signaling call throughput, with an average setup latency halved.
5
4. DESIGN AND IMPLEMENTATION
Figure 2 introduces the scenario for implementation Televoting service (or another multimedia
service) through the utilization of the JADE platform along with the OSGi framework, running on
SIP-AS.
Figure . Network Topology.
4.1. Televoting Service Lifecycle
Upon start, the OSGi Televoting service bundle loads its configuration parameters from a file
named televoto.conf and automatically creates a number of attendant agents on the JADE
platform. Each agent registers itself on IP PBX as an extension and become operational. All
registered agents are grouped in a unique number. When a call arrives at that unique number, the
IP PBX redirects it to one of the registered agents and the call is then processed. The rules for
redirection (first available, ring all, last called, etc.) depend on the IP PBX distribution. In the
proposed scenario the programed rule is first available. As described above, Televoting agent
register along IP PBX and became ready to work. Once created, they turn visible on JADE GUI
(inside MAIN container) and your control is now managed by JADE framework.
4.2. Televoting Workflow
After a call from a client is placed to a Televoting unique number, which is published on the
architecture, the initiation and maintenance of a dialogue are shown on the diagram in Figure 3
and briefly explained:

International Journal of Next
Next-Generation Networks (IJNGN) Vol.6, No.3, September 2014

1) The client places a call to a Televoting service uni
2) The IP PBX receives the call and redirects it to an attendant agent according to a configured
rule;
unique number and waits for a response;
3) One of the Televoting agents on the JADE platform receives the redirected call, signaling
client and waiting for a response.
onse.
4) The Televoting agent accepts the current call (a response to the client request made in item
1).
5) The client acknowledges the call (a response to the Televoting agent request from item 3);
6) The Televoting agent sends a media stream according to session sessio
negotiation;
7) At the end of announcement the Televoting agent hangs
The client acknowledges the Televoting agent request.
hangs-up the call, thus ending the session;
Figure 3 represents these above steps.
Figure . Network Topology.
To complement the scenario presented, we develop a graph which represents the state machine of
the Televoting service, as shown in Figure 4.
6
que


Figure . The Televoting service agent state machine.
Figure 5 presents OSGi console with JADE and televoting bundle started. As
described above,
televoting agent register along IP PBX and became ready to work. Once created, televoting agent
turns visible on JADE GUI (inside MAIN container) and your control is now managed by JADE
framework. All log messages are manipulated by an instance ins
of LOG4J bundle and recorded in a
file named televoto.log on local hard disk.
Figure . OSGi console with JADE and televoting agent initialization process.
Inside JADE container the message exchange between televoting agents and JADE architectur
components (e.g.: AMS, DF and RMA agents) can be traced by a special JADE agent called
SNIFFER Agent (messages must follow FIPA specifications). Also messages can be exchanged
between any other agents started on platform too. The number of televoting ag
defined in televoto.conf file. Figure 6 presents JADE GUI.
agent to be created is
7
tance architectural
ent


8
Figure . JADE graphical user interface and agents started on MAIN-Container.
4.3. Validation Tests
We have developed a sequence of tests to validate integration of JADE/OSGi as a SIP-AS. The
parameters observed in the experiment are: jitter, variation of jitter, packet loss and CPU load of
the SIP-AS. To support this scenario, the equipment presented in Figure 2 and detailed in Table 1
below, were used.
Table . Test equipment’s specification.
Item Equipment
SIP-AS, Client
Simulator
Dual Intel Xeon with 4 core/processor, 20 GB RAM, Intel
Gigabit Ethernet, OS Linux Server 12.04 x64
IP PBX Single Intel Xeon with 4 core, 8GB RAM, Intel Gigabit
Ethernet, ELASTIX Custom Distro
Switch H3C-2928 24 ports Gigabit Ethernet
Network
Analyzer
Intel Core2 Duo, 4GB RAM, Atheros Gigabit Ethernet, OS
Microsoft Windows XP SP3
In each test, the client simulator performed a load of calls to its counterpart in the televoting
service (a client for each service agent). This load of calls was parameterized in the configuration
file of client simulator such that it is executed one or more times, depending on the amount of
redials parameter set.
With this approach, we identified the capacity of the SIP-AS in handling calls faster without the
need of integration with the IMS Core. During testing, all network traffic was captured by the
Analyzer for subsequent analysis.
The methodology used in the tests was developed according to the following profile: (a) the
whole SIP-AS infrastructure is initialized; (b) the packet capture is initialized in the
WIRESHARK; (c) the client Simulator is initialized, running 100 concurrent calls to the


Televoting service and (d) at the end, the entire environment is shutdown. In each test, the client
Simulator is reconfigured to generate additional concurrent calls as shown in Table 2, up to a total
of 1000 calls, in order to identify its impact with respect to jitter and jitter variation. Note that the
average value of jitter, as well as your variation (MAXDELTA), increases as the load of
concurrent calls grows.
In the jitter column, we observe that even with a high load of concurrent calls, the values remain
at an acceptable threshold. However, the same does not occur with the MAXDELTA values,
because the higher the value presented more occurrences of gaps in the audio message from the
Televoting were perceived on the Wireshark RTP-analysis.
9
Table . Mean values of Jitter and its MAXDELTA.
Concurrent
Calls
Jitter (ms) MAXDELTA (ms)
100 0,72 111,80
200 1,74 213,07
300 3,03 354,79
400 3,96 401,76
500 4,94 485,36
600 5,93 522,16
700 7,29 629,50
800 9,71 784,57
900 9,91 716,44
1000 11,98 853,66
About CPU usage analysis, Figure 7 presents its load average over the tests. The minimum and
maximum load values were between 60 and 80 percent. When added more processing resources
to SIP-AS, the average value was reduced by half, leaving around 30 to 35 percent.
Figure . Average of CPU Load during tests.


The values of packet loss measured were insignificant in all tests and did not influence the
communication on media plan of televoting service. As seen in Figure 8, the time spent to attend
the n concurrent calls in Televoting service presented a behavior near the linearity as number of
concurrent calls grows.
Figure . Time interval for handling n calls in seconds.
Figure 9 depicts the average of calls handled per second on televoting servic
increase in average attendance between 100 and 400 concurrent calls per second on system. From
this point, this value remains near the stability between 30 and 32 calls handled.
Figure . Time interval for handling
service. We observe an
n calls in seconds.
We believe this is a point that the service needs to be scheduled to a second SIP-SIP
AS because we
perceive the existence of gaps (hardly noticeable) on audio message transmitted with the same
number of concurrent calls. The gaps occur due to the value of jitter j
variation to be very high, as
shown in Table 2.
10
e. itter


11
5. CONCLUSIONS
In this paper, we present the design and the development of a new service-oriented architectural
framework to support the provisioning of IP Multimedia as a service. Our approach differs from
the previous works by its functional components based on the combination of the OSGi and
JADE technologies.
As proof of concept, a prototype of an innovative Televoting service is built and deployed in a
real testbed. Results indicate that this architecture is able to support in a scalable way concurrent
calls. It allows the calls processing by service offered in PSTN Intelligent Network (IN). When
the server CPU consumption is high, it can attenuate with the vertical scaling (i.e. scale up by
provisioning of more hardware resources in a local datacenter shown in subsection 4.3) or
horizontal (i.e. scale out, distribution of the service JADE/OSGi to the cloud)
.
For future work, we plan to program the system for supporting dynamic load balancing as well as
the mobility of service agents and integration with OpenIMS Core.
REFERENCES
[1] Bellifemine, Fabio; Caire, Giovanni; Greenwood, Dominic. Developing MultiAgent Systems with
JADE. John Wiley Sons Ltd. Inglaterra. 2007.
[2] OSGi Alliance. OSGi Service Platform: Core Specification, Release 4, Version 4.2. Technical report,
2009. Disponível em http://www.osgi.org/download/r4v42/r4.core.pdf.
[3] Salchow Jr., Ken. Introduction to the IP Multimedia Subsystem (IMS): IMS Basic Concepts and
Terminology. Available in http://www.f5.com/pdf/white-papers/ims-introduction-wp.pdf
[4] Poikselkä, Miikka. Mayer, Georg. THE IMS: IP Multimedia Concepts and Services. 3 ed. John Wiley
Sons, 2009.
[5] Ahson, A. Ilyas, Mohammad. IP multimedia subsystem (IMS) handbook. CRC Press, 2009.
[6] Bellifemine, Fabio; Caire, Giovanni; Poggi, A.; Rimassa, G. JADE White Paper, disponível em:
http://jade.tilab.com/papers/2003/WhitePaperJADEEXP.pdf
[7] Wooldridge, Michael. An Introduction to MultiAgent Systems. 2 ed. John Wiley Sons, 2009.
[8] Bellifemine, Fabio; Caire, Giovanni; Greenwood, Dominic. Developing MultiAgent Systems with
JADE. John Wiley Sons Ltd. Inglaterra. 2007.
[9] Ribeiro, Renato B. C., Soares, Rosane B., The Application of JADE and OSGi Technologies in the
Telecommunications Services Architecture, EUROCON 2011.
[10] Oliveira, Rafael G. et all. An Application Server Approach for Number Portability in IMS networks.
IEEE - SBrT - ITS 2010 - International Telecommunications Symposium
September 06-09, 2010, Manaus, Amazonas, Brazil.
[11] Munadi, R. et all. Design and Implementation VoIP Service on Open IMS and Asterisk Servers
Interconnected through Enum Server. International Journal of Next-Generation Networks (IJNGN)
Vol.2, No.2, June 2010.
[12] Li, K. et al. Two Exploring Experiments on IMS Service Based on SIP AS. International Journal of
Multimedia and Ubiquitous Engineering, Vol.9, No.4 (2014), pp. 375-386.
[13] Franks, Greg et al. Performance measurement and modeling of a java-based session initiation protocol
(SIP) application server. QoSA-ISARCS '11 Proceedings of the joint ACM SIGSOFT
conference -- QoSA and ACM SIGSOFT symposium -- ISARCS on Quality of software architectures
-- QoSA and architecting critical systems – ISARCS (2011). pp. 63-72.
[14] Femminella, Mauro et al. Performance Management of Java-based SIP Application Servers. 12th
IFIP/IEEE IM 2011: Mini Conference. pp 493-500.


12
Authors
Renato Benezath Cabelino Ribeiro
Graduated in Computer Science from the Fundação de Assistência e Educação (2001). He is currently
Professor of 1 and 2 degrees of Instituto Federal de Educação, Ciência e Tecnologia do Espírito Santo. Has
experience in the area of Computer Science, with emphasis in Telecommunications, acting on the following
topics: data communication networks, multimedia applications, telecommunications, JADE/OSGi and
NGN.
Magnos Martinello
(magnos@inf.ufes.br) received his B.Sc. Degree in informatics from the Federal University of Paraná
(UFPR), Brazil, his M.Sc. Degree in computer and systems engineering from the Federal University of Rio
de Janeiro (UFRJ), Brazil, and his Ph.D. degree in computer science from the Institut National
Politechnique de Toulouse (INPT), France, in 1998, 2000, and 2005, respectively. In 2008, he joined the
Department of Informatics (DI) at the Federal University of Espírito Santo (UFES), Brazil where he
currently holds an associate professor position. Since 2011, he had a Research Productivity Fellowship
granted by CNPq. His research interests include computer networks, software-defined networks, and
performance analysis.
Celso Alberto Saibel Santos
Doctor in Fondamentalle Informatique et parallelism Université Paul Sabatier Toulouse III (1999), with
thesis work developed at LAAS / CNRS; Master in Electrical Engineering from the Escola Politécnica da
Universidade de São Paulo (1994), with work in LSI / POLI / USP; Electrical Engineering from the
Universidade Federal do Espírito Santo (1991). He is currently Adjunct Professor in the Department of
Informatics (DI) of Universidade Federal do Espírito Santo (UFES). Has extensive experience guiding and
coordinating research projects and innovation, working mainly in the areas of Multimedia, Hypermedia,
and Web.
Rosane Bodart Soares
Graduated in Electrical Engineering from the Universidade Federal do Espírito Santo (1981), MS in
Electrical Engineering from the Universidade Federal do Espírito Santo (1994) and Ph.D. in Electrical
Engineering from the Universidade Federal do Espírito Santo (2003). She is currently Associate Professor
at the Universidade Federal do Espírito Santo by the Department of Electrical Engineering. Have
experience in Electrical Engineering with emphasis on Telecommunications, acting on the following
subjects: NGN, Intelligent Networks, Platform CORBA, Distributed Systems, Multimedia Networks.

AN ARCHITECTURAL FRAMEWORK FOR DELIVERING SIP-AS MULTIMEDIA SERVICES BASED ON JADE/OSGI TECHNOLOGY

More Related Content

What's hot (20)

Viewers also liked (13)

Similar to AN ARCHITECTURAL FRAMEWORK FOR DELIVERING SIP-AS MULTIMEDIA SERVICES BASED ON JADE/OSGI TECHNOLOGY (20)

More from ijngnjournal (20)

Recently uploaded (20)

AN ARCHITECTURAL FRAMEWORK FOR DELIVERING SIP-AS MULTIMEDIA SERVICES BASED ON JADE/OSGI TECHNOLOGY