By Brough Turner
Co-founder, Senior Vice President and CTO
NMS Communications
Work on the Session Initiation Protocol (SIP) began in 1996 and the first standards track specification (RFC 2543) was out in 1999. Many had high hopes that SIP, as a peer-to-peer protocol, would redefine the very nature of telecommunications. In the intervening years, there have been some deployments, some changes and a new standards track document, RFC 3261. But so far, SIP has only a relative toehold in the world of communications.
This article discusses some of the factors that have slowed SIP’s widespread adoption and how it is currently being used. It will also shed light on the evolution of SIP deployments and future prospects for peer-to-peer SIP usage on a broader scale as well as its role in IMS infrastructures.
One SIP at a time
The biggest telecom story of the past 12 years has been the global adoption of mobile phones – over 3 billion of them in just 12 years – all using traditional circuit-switched telephony. And arguably, the most interesting telephony service enhancement after mobility has come from Skype with its seamless integration of presence, instant messaging, wideband audio and video. But Skype is based on proprietary protocols, not SIP. Finally, VoIP has helped drive down the cost of international calling, but this is the result of VoIP based on MGCP, H.248 or H.323 more than SIP, at least so far.
SIP has been adopted by PBX manufacturers in recent years, but this doesn’t seem to have changed business practices at all. The IT department still buys the PBX and the telephone sets from a single vendor and then contracts with a service provider to handle calls outside the enterprise. What happened to the idea that any SIP phone would call any other SIP phone without need of a service provider, PBX vendor and anything beyond a directory service? Unfortunately very little, at least so far.
SIP is also being adopted internally by many fixed line operators as they migrate their core networks to VoIP. SIP is widely used in gateways that make bulk connections to such operators. Indeed, Skype uses SIP to connect its proprietary Skype-to-Skype service to the service providers who support SkypeIn and SkypeOut services.
Finally, SIP itself has evolved from its initial focus on peer-to-peer communications to become the centerpiece of a complex family of protocols for next-generation networks (NGNs) envisioned by telecom traditionalists and defined by the 3rd Generation Partnership Project (3GPP).
Shhh! Some might say SIP has been co-opted. In other words, with IMS, there are no direct peer-to-peer connections. Instead everything is mediated by a service provider.
What’s next for SIP?
SIP’s biggest near-term growth will be in inter-working, especially the use of “SIP trunking” to connect IP-PBXs to IP service providers.
In parallel, there will be increasing use of SIP within networks to deliver traditional telephony services. For example, in Packet Cable 2.0, CableLabs has specified SIP as a new protocol for use between subscriber phones and the cable telephony operator, so over time, the cable industry will be migrating to the use of SIP. However, for most consumers this transition will be transparent – the cable company is providing digital POTS (plain old telephone service). The fact that SIP is involved will be significant to some equipment providers, but not to most consumers.
The good news for equipment vendors is that VoIP traffic continues to grow rapidly (a recent iLocus study reports a 35 percent annual increase in VoIP traffic in 2007). So what about those billions of mobile phone users? Unfortunately, their migration to SIP-based voice telephony networks will be delayed.
SIP for mobile voice telephony delayed
The 3GPP has specified the ultimate NGN for use by mobile operators. It goes under the name IMS (IP Multimedia System) and it’s based on SIP. It’s the ultimate NGN because the other NGN standards efforts (Cable Lab’s PacketCable, ETSI’s TISPAN and 3GPP2’s MMD) have decided to evolve their specifications to align with the 3GPP’s IMS.
IMS defines a network where every handset or other attached device uses SIP to request services. The network decides whether there are enough resources to accommodate the request and, if so, reserves those resources. That means an IMS network can guarantee bandwidth, latency and other quality of service (QoS) parameters on a per-session basis for any IP session. It also means that the operator can bill on a per-session basis and charge different prices for different services if they wish.
The trouble with all NGNs, and IMS is the ultimate NGN, is complexity. Because all communication between endpoints must be negotiated with and established by centralized network facilities, an IMS network has all the complexity of the traditional telephone network and then some. So it’s taken a long time to develop the standards, and they are still evolving. Moreover, mobile operators are still searching for a killer application that can justify the expense of migrating to IMS infrastructures using SIP.
Meanwhile pre-IMS VoIP networks continue to spread as they save money for operators today, without the complexity of IMS. There are at most a few dozen actual IMS networks in existence, they are only partial implementations and most are still in trials. In every case, IMS network elements are being deployed to support a new service that’s expected to generate new revenues – services like push-to-talk over cellular (PoC), fixed-mobile convergence (FMC) and video sharing. That said, cost savings for mobile voice telephony are based on pre-IMS VoIP technology.
Circuits-over-Packets for mobile cost savings
Mobile operators see cost savings if they can share a single packet-based core network for all service offerings, so there are plenty of operators migrating in this direction. More to the point, new networks in rapidly growing markets like China and India are beginning to use packet-based backbones to interconnect conventional mobile switches and even radio base stations. But these networks are not using SIP or IMS for voice services.
The preferred approach is to preserve conventional SS7-based mobile signaling but transmit it using SIGTRAN protocols over an IP backbone and use a softswitch to control the interconnection of voice circuits over the same IP backbone. In practice, signaling and media may be carried on separate virtual networks, but they share the same IP-multi-protocol label switching (MPLS) core network. An added advantage is there are no changes to existing signaling protocols so roaming and handoffs work as always and subscriber management is untouched. Most important, there is no need to install IMS/SIP client stacks on consumer handsets, so the billions of existing handsets continue to work.
In some cases, the softswitch vendor provides an interface, i.e. a gateway, so SIP-based equipment can connect via SIP. In other cases, traditional equipment is installed with packet interfaces in place of TDM interfaces. That means SS7 call setup messages go over SCTP/ IP instead of over MTP/ TDM and voice is sent over RTP/ UDP/ IP rather than TDM, but nothing fundamental changes. It’s the PSTN over IP or more properly, a PLMN (public land mobile network) over IP.
Long term prospects for SIP
It’s still early days for IMS, so equipment volumes are small and IMS lacks the cost savings for mobile voice telephony that circuits-over-packets approaches provide. While operators see IMS as their long-term solution, they are only investing in IMS infrastructure to support new applications that have the potential to generate new revenues. We have a chicken and egg problem that’s likely to be solved in increments over 5-10 years.
Meanwhile, SIP remains viable as a peer-to-peer protocol. As the use of SIP extends to more and more PBXs, VoIP handsets, gateways and other parts of the communications infrastructure, there will be new opportunities for peer-to-peer SIP to break through. The original vision of SIP fostering a revolution in telecommunications may yet come to pass … but it will take longer than originally predicted.