Next-generation networks (NGNs) are the result of the evolution of traditional telephony networks. The Internet protocol (IP) multimedia subsystem (IMS) is the architectural cornerstone for NGNs, and is based on Internet Engineering Task Force (IETF) protocols (session initiation protocol (SIP), Diameter). IMS supports fixed-mobile convergence, and promises to be the docking station for new revenue-generating applications, in addition to other sound business benefits (for example, operating expense reduction).
IMS open-source projects, such as the ones presented by Waiting et al., are expected to play a key role in letting a worldwide developer community set its own IMS/NGN testbeds, thereby fostering service innovation and network research. By enabling developers worldwide to replicate standards-compliant network architectures, emerging developer community projects may finally change the course of SIP, which was meant to be for multimedia services what hypertext transfer protocol (HTTP) was for the Web.
The telecommunications industry has always been service-oriented [1], but it is being challenged by the evolution of the Internet into an over-the-top service platform that bypasses telecommunications control and converts its infrastructure into simple bit pipes. Current telecom service development frameworks and practices are unable to compete with Internet service players in terms of time to market, developer base, and potential users. This is why incumbent operators are wisely embracing newly created service development kits (for example, British Telecom (BT) Web21C, Orange Partner, Vodafone Betavine, Telefonica OPEN, and DT developer) that will need to compete as well with emerging service platforms (for example, iPhone software development kit (SDK), Google Android, and Symbian) in the NGN/Internet service arena.
The IMS is seen as one more opportunity (maybe the last one) for telecom operators to adapt to the changing world, find new business models, attract service developers, run their networks more cost efficiently, enable outsourcing of infrastructure and operations, and ultimately leverage Web 2.0 with key assets like quality of service (QoS) guarantees, security, identity management, and charging capabilities.
Once the above-mentioned motivation is established, the authors dedicate one section for each of the four open-source tools that include testbed results to demonstrate both the capabilities and limitations of the developed components. The anchoring-related work is the open-source IMS Core project by the Fraunhofer Institute, which in 2006 provided publicly available code for the core IMS elements: the SIP-based Call Session Control functions and the Diameter-based user database Home Subscriber Server (HSS).
First, the University of Cape Town (UCT) IMS client addresses showstopper number one (the client side), providing an IMS end-system implementation that seeks to be a platform for IMS enablers (reusable IMS service components). The client implementation is evaluated in terms of IMS registration and call setup times over different access networks (local area network (LAN), high-speed downlink packet access (HSDPA), and Edge), coming to the conclusion that without further optimization, a slow Edge wireless access network is not suitable for IMS communications.
Second, the UCT Internet protocol television (IPTV) server presents an IMS-ready IPTV streaming server capable of offering different video qualities depending on the access network characteristics. Rather than the selected quality of the IPTV stream determined by the media codec, the server is limited by the transcoding operations, and thus presents limitations in the number of concurrent channels offered. The authors point out that this limiting factor could be resolved by having the media files already transcoded for the different qualities, or by dedicated hardware when thinking about commercial deployments.
Third, the UCT video conferencing framework tackles another issue of conferencing in IMS, namely the diverse options (centralized, distributed, and hybrid) to distribute SIP signaling and the merging of media streams. The concluding remark is that, depending on the number of participants, the conferencing framework should adopt a centralized architectural approach for scalability reasons.
Fourth, the UCT policy control framework implements a key component in IMS-based NGNs, the policy-based resource controlling functions that arbitrate between application-driven QoS requests and the network’s capabilities. A policy decision function interfaces the proxy call session control function (P-CSCF) via Diameter protocol exchange and evaluates the Extensible Markup Language (XML)-based policies that enable the operator to have control over network usage constraints and QoS classes. Policy decisions are sent to the policy enforcement point that applies the policies by gating the user traffic flows and packet-marking operations. The testbed validation of having this tight control over the SIP sessions concludes that an IMS core with QoS provisioning may suffer from unacceptable session setup delays.
In summary, the paper is a good practical overview of IMS open-source components, enabling many research opportunities. I would be more careful when extrapolating to actual commercial environments, where hardware-specific carrier-grade equipment overcomes the execution delays pointed out in some practical evaluation parts of the paper. In particular, the conclusions for the policy-based QoS part need more refinement to support the claim that per-session policy-based QoS becomes unpractical in operational settings due to the increased delay.
In any case, these tools help in understanding the detailed operations involved in delivering IMS services, and might even be key in steering the ongoing 3GPP and European Telecommunications Standards Institute (ETSI) Telecommunications and Internet Converged Services and Protocols for Advanced Networking (TISPAN) NGN standardization activities. For instance, quick proof-of-concept prototypes of new functionalities under consideration could be easily evaluated (and potentially optimized) by the consortium members, similarly to how NS2 simulations support some standardization activities.
Based on its title, readers might be disappointed with one aspect of this paper, namely one of the key points of the motivation of open-source projects: actual development community adoption experiences. I was expecting some insight and discussions on how IMS-related open-source projects have resulted in developers worldwide going beyond simple experimentation and “selfish” use of the tools to make actual code contributions to the Internet community. From my experience, I know that the mailing lists of the core IMS project create a very valuable discussion forum, however I would like to know to what extent contributions such as branches and new components are being adopted, as opposed to simple bug reports or feature suggestions to the core open-source developers.
