ROMANO LAYOUT
2/20/07
3:39 PM
Page 152
IP MULTIMEDIA SYSTEMS (IMS) INFRASTRUCTURE AND SERVICES
A Distributed IMS Enabled Conferencing Architecture on Top of a Standard Centralized Conferencing Framework Alfonso Buono, CRIAI Consortium Salvatore Loreto, Ericsson Research Lorenzo Miniero and Simon Pietro Romano, University of Napoli Federico II
ABSTRACT In this article we present an actual implementation of a distributed conferencing framework compliant with the IP multimedia core network subsystem specification. The architecture we describe has been realized by exploiting existing achievements in the field of conferencing. More precisely, starting from the IETF centralized conferencing (XCON) framework and based on an open source XCON implementation provided by our research group, we devised a distributed conferencing solution that was implemented as an overlay network of centralized conferencing clouds. The article’s goal is to provide the reader with useful information about our experience with IMS implementation and deployment. We first describe our architecture from a high level design perspective and subsequently analyze it in detail by highlighting the most notable implementation choices.
INTRODUCTION The IP multimedia subsystem (IMS) is a standardized next generation networking (NGN) architecture conceived for telecom operators willing to provide advanced services on top of both mobile and fixed networks. It makes extended use of voice over IP (VoIP) technologies based on a 3GPP implementation of SIP (Session Initiation Protocol). Heterogeneous devices are supported. Users should be able to ubiquitously exploit the entire portfolio of available services, which entails support for roaming as well as for flexible and transparent adaptation to context changes. To achieve the aforementioned goals, IMS exploits open standard IP protocols. Using this perspective, IMS is a very good example of an effort to merge the two main telecommunication agoras currently available: the Internet on the
152
0163-6804/07/$20.00 © 2007 IEEE
one hand and the cellular telecommunication infrastructure on the other. Due to its recent invention, the IMS architecture currently is far from reaching a steady state with respect to the complete definition of the overall infrastructure and related standards. Furthermore, to date only a few early trials and deployments of the architecture are underway. This means a number of open issues still must be faced, both at the infrastructure and at the service levels. The goal of this article is to contribute to the solution of some of the previously mentioned issues, with special regard to the need for actual implementation of IMS architecture and services. We focus on the implementation of a distributed conferencing framework, compliant with the IMS specification, and capable of providing conferencing facilities in conjunction with session management and floor control. The architecture we present is the outcome of our latest efforts in the field of conferencing. As explained in the article, based on our previous work on centralized conferencing, we recently moved the focus towards the management of conference information in a distributed environment. The distributed architecture we present was conceived as an extension to a standard centralized conferencing framework [1], under definition inside the IETF XCON working group. Basically, it envisages the construction of an overlay network acting as glue between a number of centralized conferencing “islands.” We discuss the ideas that inspired our work and subsequently provide information about our design and implementation choices. The article is structured as follows. We provide the background of our work by introducing the reference context, as well as by briefly describing our previous work and achievements in the field of centralized conferencing. The requirements for an IMS-compliant distributed
IEEE Communications Magazine • March 2007
ROMANO LAYOUT
2/20/07
3:39 PM
Page 153
architecture are identified. We describe our proposal for a distributed conferencing framework that was conceived at the outset by taking into account the identified requirements. Implementation details are illustrated, and later we deal with related work. Finally, we provide some concluding remarks, together with information about our future work.
CONFIANCE: AN OPEN SOURCE IMS ENABLED CENTRALIZED CONFERENCING ARCHITECTURE We recently presented, as a collaboration activity involving the University of Naples and Nomadic Lab, an open source conferencing framework that we call CONFIANCE (CONFerencing IMS-enabled architecture for nextgeneration communication experience). The framework is intended to be compliant with the IMS specification by taking into account ongoing standardization efforts inside the various active international bodies (IETF, 3GPP, OMA, etc.). As explained in the next section, compatibility with the IMS architecture is realized through a mapping between the logical IMS entities and several real-world components. To offer advanced conferencing capabilities, great efforts were made to develop an actual implementation of the centralized conferencing framework as proposed by the IETF working group XCON, from which we took inspiration. We worked on both the server side (focus) and the client side (participant and administrator), as well as on the communication protocols between them, namely the Conference Control Protocol (CCP) and the Binary Floor Control Protocol (BFCP). CCP handles session management, and BFCP handles moderation aspects of the framework, to enable coordinated access to the set of resources it offers. Although BFCP already was completely specified [2] by the XCON WG, no agreement yet exists on a specific candidate for the conference control protocol. This led us to implement a brand new protocol (called XCON Scheduler) to handle the basic session management functionality the framework must provide (e.g., creating new conferences or getting information about them). The reader can refer to the official project web page [3] to obtain information about our implementation choices. The framework currently provides many advanced conferencing features, with complete audio support and basic video switching (one speaker at a time).
TOWARDS AN IMS COMPLIANT DISTRIBUTED CONFERENCING FRAMEWORK In this section we help the reader position our work. For the sake of conciseness, we do not provide details about the standard IMS components, as we assume that the reader already is familiar with the IMS architecture. Hence, in the
IEEE Communications Magazine • March 2007
following subsections, first we provide information about the proper mapping between our architecture components and the corresponding IMS logical entities in the centralized conferencing scenario. Afterwards, we move to the distributed scenario, which requires the introduction of a new logical IMS element, capable of distributing conference information among several IMS core networks. For both the previously mentioned scenarios, we also describe how to replace the identified logical entities with real-world components.
CENTRALIZED CONFERENCING SCENARIO To date, the 3GPP specified the required functionality for a centralized conferencing framework within the IMS [4]. The cited document defines the following conferencing logical entities: •Conference participant: It must be implemented in the user equipment (UE), which must be SIP-, CCP- and BFCP-compliant to correctly support the floor participant and/or floor chair role. All messages involving the UE (in either direction) traverse the P-CSCF (proxy-call session control function), which represents the access point to the IMS network. •Focus, floor control server, conference notification service, conference policy server and media policy server elements are logically colocated within a SIP AS/MRFC (application server/media resource function control) component. All SIP messages involving the UE traverse an allocated S-CSCF (serving-call session control function). The S-CSCF inspects each message and determines the correct AS to which it must be forwarded for further processing. In our scenario, an AS is a conference focus. The MRFC controls media stream resources in the MRFP (media resource function processing) using information from both the AS and the S-CSCF. •Media mixer is hosted in the MRFP. According to the identified logical entities, we can replace the IMS elements with real-world components. In our architecture, the open source community provides some of these components. Other entities either were developed from scratch or based on open source artifacts that were appropriately extended to meet our architecture requirements. As described in Fig. 1, we replaced the PCSCF with a SIP Proxy server called OpenSER [5]. The S-CSCF and HSS (home subscriber server) elements were realized by exploiting an open source PBX called Asterisk [6]. Asterisk provided us with many of the required IMS functions. In fact, an enhanced version of an Asterisk component called MeetMe, capable of managing conferences, plays the role of the SIP AS in our architecture. A few ad-hoc modified Asterisk modules that provide media management, streaming, and floor control play the roles of the MRFC and MRFP components. Finally, we replaced the UE with a SIP client called Minisip [7], capable of handling both CCP and BFCP protocol messages.
In our architecture, the open source community provides some of these components. Other entities either were developed from scratch or based on open source artifacts that were appropriately extended to meet our architecture requirements.
DISTRIBUTED CONFERENCING SCENARIO This section describes the design and implementation of a novel architecture for distributed conferencing, which was the main subject of our
153
ROMANO LAYOUT
2/20/07
3:39 PM
Page 154
Legacy PLMN C, D,
SIP-AS Sh
OpenSER
ISC
So
Mb
Si
BPCP Mb
ISC
MRFP
Mw Mw
P-CSCF
MRFC Mr
ISC
P-CSCF
OpenSER
IM-SSF Gc, Gr
OSA-SCS
S-CSCF
HSS
Cx Dx
Mw Cx Mw Mw
Mi Mk
Mm
SGW
SLF
Dx
Mg Mj
I-CSCF
MGCF
BGCF
Mn Mb
Non-IMS IP PDN
MGW
■ Figure 1. Conferencing and IMS: proposed mapping. work after we completed the centralized framework. Under the assumption that all conference participants refer to the focus in the home network of the conference initiator, the IMS centralized conferencing solution also works in the scenario where the conference participants belong to networks owned by different telecom operators. However, in a fully distributed scenario, the previous case should be dealt with by providing each involved network with a specific focus that manages the associated local users who subscribed to the conference. The focus in the conference initiator’s home network must provide all other focus entities with up-to-date conference information. These foreign focus entities play a two-fold role. On the one hand, they act as a regular conference focus for the participants belonging to their underlying managed network; on the other, they appear as normal participants to the focus in the conference initiator’s home network. This requires a dedicated communication channel among the distributed focus entities that are participating in the conference. This channel is used to exchange conference information, as well as to manage synchronization issues. To make available the previously mentioned functionality, we modified the standard behavior of the IMS SIP-AS/MRFC component by providing it with the capability to manage the distribution of messages to neighboring IMS core networks. In the mapping process (shown in Fig. 2), we replaced the UE with an integrated client capable of offering both media streaming and conference management/monitoring functionality. As we explain in the next section, this client actually was implemented by enabling the previously
154
mentioned BFCP-enabled version of Minisip to interwork with an open source XMPP (Extensible Messaging and Presence Protocol [8]) client called Spark [9]. Spark, in turn, was modified to be capable of handling CCP protocol messages. We also replaced the standard IMS SIP-AS element with a component we implemented ourselves that is made of two parts. One part is the previously mentioned enhanced version of the Asterisk MeetMe application; the other is an adhoc extended version of an instant messaging server called Wildfire [10]. Wildfire plays a key role in the distributed scenario, as it enables the effective management of the required communication procedures among heterogeneous IMS core networks.
DCON: A PROPOSAL FOR A DISTRIBUTED CONFERENCING FRAMEWORK The distributed conferencing architecture we defined was conceived to be highly reliable and scalable. It is called DCON, for distributed conferencing, but at the same time explicitly recalling the previously mentioned XCON model. DCON is based on the idea that a distributed conference can be established by appropriately orchestrating the operation of a set of XCON focus elements, each in charge of managing a certain number of participants distributed across a geographical network. Interaction between each participant and the corresponding conference focus is based on the standard XCON framework, whereas interfocus interaction was completely defined and specified by us. We propose the adoption of a suitable set of protocols that are complementary to the call signaling protocols and are required for supporting advanced conferencing applications.
FRAMEWORK REQUIREMENTS To build distributed conferencing on top of an already available centralized conferencing framework, we introduced two major functions: •A coordination level among conference focus entities •A way to effectively distribute conference state information The coordination level is required to manage a distributed conference for its entire lifecycle. For example, after a user decides to create a new conference, the corresponding conference focus must distribute conference information to all other foci, to enable other potential participants to retrieve the required data and possibly, to subscribe to the event. We assume that all the operations required inside a single conference “realm” are managed via the XCON protocols and interfaces, as implemented in Confiance. Hence, each single realm continues to be based on a star-topology graph for all that concerns the call signaling part. The various available stars are then connected through an upper layer mesh-based topology providing interfocus communication. As to the second point mentioned previously, it seems clear that a way to propagate informa-
IEEE Communications Magazine • March 2007
ROMANO LAYOUT
2/20/07
3:39 PM
Page 155
A fully meshed
Client C - spark / MiniSip
overlay network in which each node
Access network
plays the role of the focus element of an
IMS core network Client C home network
XCON island represents the top layer. The lower
S2S channel
S2S channel
layer, in turn, IP backbone
IP backbone
is characterized by a Client D - spark / MiniSip
Client A - spark / MiniSip Access network
S2S channel
S2S channel IMS core network Client A home network
Access network
hub is represented by the focus element) and is fully
Client B - spark / MiniSip Access network
IMS core network Client D home network
star topology (in which the central
S2S channel
IP backbone
S2S channel
Access network Client E - spark / MiniSip
compliant with the XCON specification.
■ Figure 2. Distributed conferencing and IMS: proposed mapping. tion about conferences is required when switching the view from a centralized to a distributed perspective. Indeed, whenever a new conference is created (or an active conference changes its state) this event must be communicated to all interested (or active) participants. Given the intrinsic nature of the distributed framework, the actual flow of information always foresees the interaction among conference focus entities for both conference information exchange and for state change notifications. Conference state propagation can occur in a number of alternative ways. For example, each focus might flood the received information across the inter-focus communication mesh, thus guaranteeing that potential participants belonging to heterogeneous islands are reached. In such a case, focus entities are stateful, that is, each stores information about current sessions and forwards such information to all peering entities to bring them up-to-date with respect to available conference sessions. On the other hand, a distributed repository might be employed to store conference information: focus entities would access this repository, both to publish (either upon creation of a new conference or to notify a change in the state of an active conference) and to retrieve information about active conferences (e.g., when a new participant wants to access the list of ongoing/scheduled conference sessions). In this last case, focus entities are stateless.
FRAMEWORK DESIGN DCON was conceived as a large scale evolution of our Confiance framework. We deploy our architecture on top of a two-layer network topology (Fig. 3). A fully meshed overlay network in which each node plays the role of the focus element of an XCON island represents
IEEE Communications Magazine • March 2007
the top layer. The lower layer, in turn, is characterized by a star topology (in which the central hub is represented by the focus element) and is fully compliant with the XCON specification. In the DCON scenario, communication among different islands becomes of paramount importance. To the purpose, we are adopting the so-called S2S (server to server) module of the XMPP protocol. XMPP was standardized by the IETF as the candidate protocol to support instant messaging, e-presence, and generic request-response services. XMPP seems to us to be the ideal communication method among DCON focus entities. In the following section we present a prototype solution based on the stateful approach. A stateless implementation of the overall architecture currently is being performed by our research group and is not the subject of this article.
DCON AND THE IMS ARCHITECTURE SIP definitely represents one of the fundamental choices upon which to base the IMS architecture. In fact, it is taken for granted that any IMS-compliant implementation of a service should rely on SIP for everything concerning the establishment of a communication session. During the design phase of the DCON framework, we departed from this guideline, as we assumed that interaction between conferencing clients and the conference management entities can exploit either SIP or an alternative protocol. More precisely, we also gave our clients the capability to interact with the framework by means of an instant messaging paradigm. Hence, a DCON conference can be created and managed through triggers issued either from an enhanced SIP client (i.e., a client supporting SIP as well as BFCP and a suitable conference control protocol) or from an instant messaging client that can optionally be associated with the
155
ROMANO LAYOUT
2/20/07
3:39 PM
Page 156
Inside DCON, communication between the legacy
XMPP SIP (BFCP enabled) (BFCP enabled)
Confiance modules and the newly created distribution components occurs on the basis of an asynchronous
S2S server to server channel
paradigm in which a number of events are generated by Confiance modules
XMPP (BFCP enabled) SIP (BFCP enabled)
the XCON island they currently
S2S server to server channel
SIP (BFCP enabled) XMPP (BFCP enabled)
S2S server to server channel
whenever something relevant occurs in
S2S server to server channel
S2S server to server channel
supervise. SIP XMPP (BFCP enabled) (BFCP enabled)
■ Figure 3. DCON design: the stateful approach.
SIP client itself. The final scenario we envisage is one in which a single client integrates both the enhanced SIP functionality and the instant message capabilities.
DCON IMPLEMENTATION In this section we provide information about the current implementation of the DCON framework. We now focus on a prototype that implements the DCON basic functionality by assuming a stateful scenario. Source code and documentation for this prototype can be found at the official project web page on sourceforge [11]. We also note that the current implementation focuses only on the distribution of conference information, and does not provide comprehensive conference management functionality. Figure 4 depicts the main implementation choices of the DCON architecture. The lower box of the picture presents the logical view of the server side, integrating our Asterisk based Confiance implementation of the XCON focus (left-hand side) with a brand new module specifically conceived for the SPreAding of Conference Events (which we call SPACE). SPACE is realized as a plug-in for Wildfire, a popular open source instant messaging server. We chose Wildfire, as it foresees the possibility of interacting with an Asterisk server through an ad hoc module called Asterisk-IM. SPACE actually represents a key component of the architecture, since it enables inter-focus communication through the exchange of con-
156
ference information. Interaction between Wildfire and Asterisk occurs through the AsteriskIM plug-in, making use of the Asterisk Manager Interface Protocol (AMP). Inside DCON, communication between the legacy Confiance modules and the newly created distribution components occurs on the basis of an asynchronous paradigm in which a number of events are generated by Confiance modules whenever something relevant occurs in the XCON island they currently supervise. The upper box presents a view of the client side that logically can be viewed as a single, integrated entity capable of interacting with the framework by means of either SIP or instant messaging. As explained previously, the SIP part of the client was realized through the open source Minisip softphone, appropriately modified to support both the BFCP and the CCP protocols. For the IM client, we chose to adopt Spark, an open source cross-platform client using the XMPP protocol. We added the capability to interact with the DCON platform through an ad hoc created plug-in (SpaceClient in the figure) to Spark.
THE SPACE MODULE FOR INTERFOCUS INTERACTION The SPACE module was realized to enable DCON functionality both on the server and on the client side. The following explains the respective roles in further detail.
IEEE Communications Magazine • March 2007
ROMANO LAYOUT
2/20/07
3:39 PM
Page 157
SPACE Server Side — On the server side, we were required to introduce support for the following new features: • Core functionality of the DCON focus • Focus discovery • Event distribution Regarding the first new feature, we concentrated our efforts on the integration between the Asterisk-based Confiance server and the newly introduced DCON Wildfire component. To offer a richer experience to users, we also implemented a direct communication channel with the Confiance XCON scheduler for session management. Both of the previously mentioned features rely on the existence of a repository, which we implemented from scratch. The repository contains relevant information regarding active and registered DCON conferences. With the discovery of other DCON focus entities, we chose to implement the new feature based on a client-driven IM presence service. Whenever the first client becomes active in a DCON cloud, the related focus entity opens an S2S channel to all other focus entities it is aware of, thus triggering the distribution phase. Finally, with regard to the distribution of conference information, after the S2S channel was established as previously explained, the focus entities at both edges of the channel begin sending each other their stored data. Triggers coming from the associated Confiance module generate data flowing along the S2S channel. As depicted in Fig. 5, each time something relevant occurs in an XCON cloud, Confiance raises an event through the Asterisk manager interface. The event is then intercepted by the SPACE plug-in that in addition to alerting all the associated SPACE clients, forwards it across all established outgoing sessions. SPACE Client Side — On the client’s side, we introduced the following new functionality: • Retrieval and visualization of conference information • Support for the creation of a new DCON conference • Support for the joining of an existing DCON conference In the following section we briefly touch on the new functionality, with special reference to the IM-based side of the client DCON component. We do not mention the SIP-initiated scenarios as they do not present new facets compared to a centralized conference that we already addressed within the Confiance framework. In the visualization of conference information, we provided the Spark client with a dedicated panel showing the details (e.g., IP address of the reference focus, conference identification number, conference status) of both the active and scheduled DCON conferences. Information viewed by the user is retrieved through the exchanging of XMPP messages across a dedicated channel connecting the client to its reference focus. As described in the previous section, upon activation of the first connection in an XCON cloud, the focus opens the S2S channels to the peering entities and immediately thereafter sends the retrieved conference infor-
IEEE Communications Magazine • March 2007
Integrated SIP/IM client
Spark
Enhanced MiniSip
SpaceClient
Manager
Asterisk-IM
CONFIANCE Space
Wildfire
Scheduler Asterisk
Focus s2s
■ Figure 4. Implementation of the DCON architecture.
mation to the client itself. Even when an active connection between the client and the focus is already in place, it is still possible for the client to asynchronously contact its reference focus to have up-to-date information about available conferences. Figure 6 shows the relevant information flow with regard to the IM-triggered creation of a new DCON conference. The conference is created by sending a CCP message (message number 1 in the picture) to the Confiance scheduler across a direct connection between the SPACE client and the Confiance focus. Upon receiving the message, the scheduler component instantiates a new conference, whose information is sent back to the client (message number 2 in the picture). We note that each IM client relies on a specific softphone to place conference calls. Currently, this is required because the existing implementation of the framework does not provide a client component integrating enhanced SIP functionality with IM-based capabilities. After the client is notified of the successful creation of a conference, it sends an XMPP message to the Wildfire Asterisk-IM module (message number 3 in the picture) to assess whether it is ready to participate in the newly created conference. The Asterisk-IM module, in turn, sends a message to the Asterisk manager interface (message number 4 in the picture) which originates a new call — for example, a SIP INVITE message — between the conference room and the associated soft-phone (message number 5 in the picture).
157
ROMANO LAYOUT
2/20/07
3:39 PM
Page 158
To the best of our knowledge currently,
Local event storing
CONFIANCE
no integrated effort has been made that
Database Remote event storing Active focus entities
Event
Incoming session
provides a real architecture that is both IMS compliant
SPACE
Manager Event notification
Asterisk
Wildfire
S2S Outgoing session
and capable of offering distributed
■ Figure 5. The spreading of conference information.
conferencing functionality. This is due mainly to the fact that no agreedupon solution has been designated with respect to certain crucial points. Scheduler
SPACE
AsteriskIM
Asterisk
1 create 2 conf info (CCP)
3 conf info (XMPP)
4 conf info (SIP)
5 incoming call
■ Figure 6. IM-triggered creation of a DCON conference. The case of a client joining an existing conference is quite similar to conference creation, namely, when joining an existing conference, the IM client simply contacts the Asterisk-IM module to enable it to initiate a new call to the softphone associated with the client itself.
RELATED WORK The architecture we presented in this article focuses on two main features: • Compatibility with the IMS framework • Capability to offer advanced conferencing functionality in a distributed scenario Although there is a rich literature about each of the previous points when considered by itself, to the best of our knowledge currently, no integrated effort has been made that provides a real architecture that is both IMS compliant and capable of offering distributed conferencing functionality. This is due mainly to the fact that no agreed-upon solution has been designated with respect to certain crucial points, such as the choice of the most suitable paradigm for distributed conferencing, as well as its integration in the general IMS architecture. With respect to the first point, no official IETF activity has begun on the distributed scenario. Nonetheless, a few research works already have proposed moving from a centralized to a distributed per-
158
spective. This is the case, for example of [12], where the authors propose a model to extend the SIPPING (Session Initiation Proposal Investigation) approach to a distributed scenario. With respect to the cited work, we propose an implementation of the distributed framework based on an extension to the XCON approach. When compared to SIPPING, XCON can be considered to be a much more focused effort, as it just deals with conferencing, for which it introduces advanced functionality such as floor control and independence from SIP as the media signaling protocol. As further evidence, the authors of [12] do not provide any implementation of their proposed architecture. All information contained in their article was retrieved through simulations. Contact points with our work also can be found in [13], which proposes a model for scalable video conferencing, based on the dynamic delegation of the management of conference services through an object-oriented approach. The cited article conforms to a centralized paradigm for session management functionality. Another remarkable research work is the one proposed by the authors of [14], which explores the possibility of using a service oriented architecture to provide flexible and scalable conferencing facilities. However, the cited project does not seem to be oriented towards a future integration with the IMS architecture. On
IEEE Communications Magazine • March 2007
ROMANO LAYOUT
2/20/07
3:39 PM
Page 159
the IMS side, efforts already were devoted to the realization of testbeds, as in the case of [15]. In that article, the authors propose a testbed for multimedia services, with support based on the IMS specification.
CONCLUSIONS AND FUTURE WORK In this article we presented an IMS-compliant architecture offering a distributed conferencing service with enhanced functionality, such as conference scheduling facilities and conference moderation. Recently, we implemented a proof-of-concept prototype, capable of effectively supporting the creation and management of a distributed conference in a scenario involving a number of IMS-compliant core networks, interconnected through a communication channel created on an ad hoc basis. However, many challenging issues still must be faced, mainly related to two different aspects. First, the current implementation of the conferencing framework offers complete support for just one medium, namely audio. Support for video is still in its early stages; work is in full swing inside our group to move from the basic video switching functionality we currently offer to more powerful video mixing/transcoding features. Another direction deals with the implementation of the stateless scenario for distributed conferencing, as well as to the realization of a thorough experimental campaign aimed at both evaluating the performance of the overall framework and identifying its potential bottlenecks (with special regard to scalability and reliability properties).
ACKNOWLEDGMENTS The authors of the article would like to acknowledge Tobia Castaldi for his work on the distributed conferencing framework. Tobia participated in the definition of the overall architecture and was responsible for the implementation of the SPACE module for the distribution of conference information.
REFERENCES [1] M. Barnes, C. Boulton, and O. Levin, “A Framework and Data Model for Centralized Conferencing,” draft-ietfxcon-framework- 06.txt, Dec. 2006. [2] G. Camarillo, J. Ott, and K. Drage, “The Binary Floor Control Protocol (BFCP),” RFC4582, Nov. 2006. [3] CONFIANCE Project page (sourceforge); available at http://confiance.sourceforge.net [4] TS 24.147 7.1.0 Technical report, 3GPP, “3GPP, Conferencing Using the IP Multimedia (IM) Core Network (CN) Subsystem; Stage 3,” Mar. 2006. [5] “OpenSER — the Open Source SIP Server”; available at http://www.openser.org [6] “Asterisk — the Open Source PBX”; available at http:// www.asterisk.org [7] “MiniSIP — An Open Source SIP User Agent”; available at http://www.minisip.org [8] P. Saint-Andre, “Extensible Messaging and Presence Protocol (XMPP): Core,” RFC3920, Oct. 2004.
IEEE Communications Magazine • March 2007
[9] “Spark — An Open Source, cross-platform IM client”; available at http://www.igniterealtime.org/ projects/spark [10] “Wildfire — A Real Time Collaboration (RTC) server”; available at http://www.igniterealtime.org/ projects/wildfire [11] DCON Project page (sourceforge), available at http:// sourceforge.net/projects/dcon [12] Y. Cho et al., “Policy-Based Distributed Management Architecture for Large-Scale Enterprise Conferencing Service Using SIP,” IEEE JSAC, vol. 23, Oct. 2005, pp. 1934–49. [13] Z. Yang, M. Huadong, and J. Zhang, “A Dynamic Scalable Service Model for SIP-based Video Conference,” Proc. 9th Int’l. Conf. Comp., Supported Cooperative Work in Design. [14] W. Wenjun et al., “Service Oriented Architecture for VoIP Conferencing,” Int’l. J. Commun. Systems, vol. 19, 2006, pp. 445–61. [15] T. Magedanz, D. Witaszek, and K. Knuettel, “The IMS Playground @ Fokus An Open Testbed For Next Generation Network Multimedia Services,” Proc. 1st Int’l. Conf. Testbeds and Research Infrastructures for the DEvelopment of NeTworks and COMmunities (TRIDENTCOM’05), 2005.
BIOGRAPHIES ALFONSO BUONO received his degree in Computer Engineering from the University of Napoli Federico II, Italy, in 1999. He is currently a senior researcher at the Italian CRIAI research consortium. His research interests are primarily in the field of mobility, with special regard to adaptive applications and systems for mobile environments, mobile distributed systems, and location-dependent applications. He is currently involved in a research project whose main goal is the design and implementation of an IMS enabled distributed conferencing architecture. He is also leading a project to design and implement a reliable architecture for the delivery of value added services in a mobile environment.
Recently, we implemented a proof-of-concept prototype, capable of effectively supporting the creation and management of a distributed conference in a scenario involving a number of IMS-compliant core networks, interconnected through a communication channel created on an ad hoc basis.
SALVATORE LORETO is an active member of the IEEE Computer Society. He received his degree in Computer Engineering from the University of Napoli Federico II, Italy, in 1998. Since then, he has been working for Ericsson Research Labs, while also working towards his Ph.D. in the Computer Networks group at the University of Napoli. His research interests are primarily related to the multimedia area, with special focus on SIP-based applications and communication protocols. He is involved in Ericsson’s standardization activities, both at the IETF and inside the Open Mobile Alliance. He also is an active member of the ACM. LORENZO MINIERO received his degree in Computer Engineering from the University of Napoli Federico II, Italy, in 2006. Since then he is actively working with the Computer Science Department of the University of Napoli. His research interests mostly focus on network real-time applications, such as IP-based telephony, multimedia conferencing, and communication protocols. He is currently involved in the definition and implementation of a conferencing framework that takes into account ongoing standardization efforts inside several active international bodies. S IMON P IETRO R OMANO (
[email protected]) is member of the IEEE Computer Society. He received his degree in Computer Engineering from the University of Napoli Federico II, Italy, in 1998. He obtained a Ph.D. degree in Computer Networks in 2001. He is currently an Assistant Professor in the Computer Science Department of the University of Napoli. His research interests primarily are in the field of networking, with special regard to QoS-enabled multimedia applications, network security, and autonomic network management. Currently , he is involved in a number of research projects, where the main objective is the design and implementation of effective solutions for the provision of services with quality assurance over premium IP networks. He is also a member of the ACM.
159
