VoIP has been found to be capable of delivering effective voice calls. However, current solutions require careful network engineering at present to provide sustained voice quality. With a number of large players very active in the development of equipment, the performance of the technology continues to improve rapidly and unit costs continue to decline. On a cautionary note this has been driven by a large amount of predatory pricing and confusion between price and cost. There is little doubt that these aspects are not sustainable in the future. With inherent capabilities for the support of multimedia, VoIP offers a rich fabric upon which more advanced applications can be built. Through the use of the ubiquitous networking offered by suitable IP network solutions, such applications can be readily built without the need for complex network infrastructure upgrades. The current challenges remain in the area of achieving the right functionality from both the IP network and a controlling application layer that recognises the stakeholders identified and described in this chapter. However, VoIP needs to be set in the wider context of multimedia broadband networks - including third generation wireless - as it becomes less viable to contemplate single-service-type networks in the future. In other words, VoIP enables, and should be considered part of, a new generation network architecture and industry that reflects the separation of multimedia bearer services from true content. In this context, VoIP is only likely to reach its full potential where the networking concerns have been appropriately separated from content and application concerns.

In this chapter, VoIP is just one of a number of potentially exciting applications for broadband access, which is now much more widely available with the roll-out of DSL. However, even when VoIP is considered in isolation, the nature of the access network is such that considerable care has to be taken over the network design if an efficient service of acceptable quality is to be delivered. When VoIP is just one of a mix of different services, then the challenges become much greater. Although the distribution of VoIP in customer premises in the business environment is not likely to be a major issue (apart from the firewall and NAT-related issues discussed in Chapter 5), in the residential environment it poses some significantly greater problems that have not yet been fully solved.

This chapter aims to explore what standards are, why they are required for realising voice over IP (VoIP), and how they are created. The trite answer to why we need standards for VoIP is that it is based upon IP and IP is a standard and so you cannot have VoIP without it. However, there are numerous, and potentially less obvious, reasons why standards are essential to VoIP. Some of these are a consequence of the natural separation of transport and application that occurs in VoIP systems; others derive from the establishment of any voice service, especially one that crosses the boundaries between the domains of different service or network providers and so encounters differences in administrative policy - in other words, the real world is much more complex than the laboratory bench.

A device that sits between two different types of network technology is called a 'gateway'. A common gateway for telephony sits between the current TDM domain and the emerging packet domain, such as an IP network. In principle, gateways could connect directly either to an analogue telephone, to a PBX, or to a TDM telephony switch on one side, and to an ATM, IP or MPLS network on the other side. Gateways can also connect two packet domains. Even a device that provided transcoding between two compression algorithms, but otherwise maintained the same transport, could be considered to be a gateway. In the context of the media gateway control protocols discussed in this chapter, gateways may also include devices that provide services to audiovisual conferences, such as MCUs and IVR boxes.

This chapter discusses why both names and addresses are necessary in a VoIP environment, and described the part that both play in the establishment of calls. It has highlighted the importance of names and addresses being unique within the context of the system in which they are used. The existing naming scheme based on E.164 numbers has been described and it has been explained why this will continue to be required for many VoIP applications. There are, however, a number of alternative options that may become available in the future and a method of handling some of these has been identified in outline.

Current Internet applications, such as e-mail and Web-browsing, are client/server based and do not require session set-up or control. They work very effectively on best-effort IP networks. In the future, it is envisaged that both fixed and mobile services will be provided by a QoS-enabled all-IP core network. The next generation of peer-to-peer style applications, however, require advanced functionality in terms of: user location by name; media negotiation; and session renegotiation on handover, and the like. This functionality is analogous to call control and intelligent network-based call treatment in the PSTN world, except that it is many times richer and potentially more complex. SIP is an excellent candidate for any peer-to-peer applications capable of communication over IP because it is lightweight, flexible, scalable, extensible and programmable. This chapter has shown how the concept of a SIP proxy server can be used to provide personal mobility and allow complicated 'intelligent' services to be delivered. This has included familiar PSTN services, such as call forwarding, as well as new mobile multimedia services. SIP offers a great opportunity for both network and service providers, such as ISPs, to create carefully tailored services, control the process of session initiation to enable higher quality and more reliable services, and enable the use of chargeable network facilities, such as archivers, media codecs and the like.

Voice is an integral element of many of the emerging breed of conversational IP applications found on the Internet. This chapter examines the opportunities to provide these types of application at scale, and considers some of the difficulties in translating what may be interesting toys on the lab bench into genuinely useful public-scale communication services. It characterises early VoIP network service solutions, charting their evolution towards contemporary services and considers some of the types of application that may be possible in the future.

This chapter considers the design and deployment of an application solution that spans both existing TDM networks and emerging VoIP networks. The first Recorded Information Distribution Equipment (RIDE) system was deployed by BT in the late 1980s to provide economic mass access to recorded announcements. The system was designed for excellent availability for applications with a high customer profile, generating a very large number of calls. Services deployed on RIDE include announcements for national code changes, timeline (the 'speaking clock') and televoting.

VoIP is a maturing, yet still emerging, technology, and its products are therefore both relatively new in the market-place and continually evolving. Such a rapidly changing technological environment for service providers, equipment manufacturers and end users is compounded by continually developing regulatory policy. Any potential purchaser of a new technology therefore has the dilemma that equipment may become rapidly dated by new developments or lack of features subsequently identified as essential requirements for commercial applications. As a consequence, establishing the precise capabilities of specific offerings at a given point in time is essential for anyone with a serious interest in the VoIP industry. This chapter describes an approach to evaluating VoIP solutions, the problems associated with the testing of VoIP networks and components, and the tools and techniques employed to do so. It outlines the capabilities of the VoIP technology evaluation test bed, which is administered by BTexact Technologies at Adastral Park, and includes some typical results obtained using perceptual analysis measurement and IP net work simulation technologies to assess voice quality.