September 2001

Creating Flexible IP Networks With SIP

BY DWIGHT IRVING

[ Go Right To: SIP Application Servers ]

Adopted in 1999 by the Internet Engineering Task Force (IETF), SIP provides for the seamless transmission of voice, fax, and data across IP and traditional telephone networks. The IETF defines SIP as "a text-based protocol, similar to HTTP and SMTP, for initiating interactive communication sessions between users. Such sessions include voice, video, chat, interactive games, and virtual reality."

SIP is used to establish peer-to-peer media sessions on an IP network including Internet telephony, conferencing, instant messaging, and unified messaging. It uses less bandwidth than H.323 because it requires fewer and smaller messages to set up and tear down a call. SIP has become a standard IETF protocol for signaling in third-generation mobile networking because it is able to initiate, modify, and terminate calls on the network and because it allows end users to become "terminal independent."

Industry pundits and the trade press feel that the time for SIP has come (search www.tmcnet.com for examples). In addition, some of the leading telecom equipment manufacturers, including Ericsson, Cisco, and Nortel, are developing SIP-compatible infrastructures.

The Power Of SIP
Users on a SIP network may have a number of devices or IP addresses on the network, and proxy, redirect, and location servers provide the functionality to unite these devices to provide a single address for unified communications/messaging.

According to the IETF RFC 2543 specification, SIP addresses the following criteria for establishing and terminating multimedia communications:

User location -- determination of the device to be used for communication.

User capabilities -- determination of the media and media parameters to be used.

User availability -- determination of the willingness of the called party to engage in communications.

Call setup -- establishment of call parameters at both called and calling party (alerting).

Call handling -- including transfer and termination of calls.

SIP includes an interesting concept called "presence." Users can make a SIP connection between two endpoints, but the interesting element is that each endpoint can be any number of devices including: cell phones, pagers, desktop phones, Internet phones, a Blackberry wireless e-mail device, or other PDA. The SIP environment determines where the user is, how willing the user is to accept communications, the terminal device's media capabilities, and which devices have priority for connection attempts. SIP may try to connect with each device one at a time or send an "invite" to multiple devices at the same time (multicast). Presence services range from the simple follow-me and instant-messaging services to highly advanced community gaming concepts.

A SIP network contains a proxy server, which is analogous to a router. Common proxy services include basic address translation, redirect, multicast, load balancing, registration, and authorization. Most proxy servers have the ability to modify and configure existing services or provide additional services through HTML, XML, or Java script. This capability allows existing services to be easily reconfigured and new services to be defined and easily added on a per-user basis. The proxy server takes the SIP call, determines the correct address for the recipient, and sends the invite. A location server finds the user and the specified device. Users can designate different devices for different times of the day. For instance, Bob is in a meeting in the morning and elects to receive notification via his Blackberry, because it offers a vibrating alert that won't disrupt the meeting. Once the meeting is over, Bob selected his cell phone as the message destination device, and so on.

The advantages of a scalable, easily deployable SIP network are straightforward, and the application and service opportunities are numerous. Read on for more a more in-depth technical look at SIP networking.

SIP In The Next-Gen Network
To better understand what logical service components are connected and communicated through SIP, we need to look at how these functional elements are divided or partitioned.

Traditional and alternative next-generation network architectures may differ in where functionality is partitioned, but SIP usage is a common element. Our distributed, or decomposed, gateway architecture, consisting of a signaling gateway, media gateway controller, media gateway, and media server (ftp://ftp.dialogic.com/www/pdf/7299.pdf), uses SIP as an access protocol between terminals and SIP-enabled NSPs (such as Level 3 Communications, WorldCom, etc.). SIP is also useful as a lightweight protocol to connect the media stream between banks of media gateways and media servers.

Alternative next-generation network architecture models may elect to use the SIP proxy server as a replacement for both the signaling gateway and media gateway controller. In this model, functionality partitioning is more similar to current IP networking services, and therefore more familiar to the hordes of network application developers than traditional telephony models. The SIP proxy server -- or different flavors of proxy servers -- takes on the role of router, name server, load-balancing switch, and domain access validation server.

Four flavors of proxy server, as identified by dynamicsoft that pertain to this discussion include: Access, Edge, Firewall Control, and Core. In brief, the Access proxy server resides at a local ASP, CSP, or enterprise and provides user-side access to the SIP NSP. The Edge proxy server provides entry into an NSP and authentication, account, and denial of service (DoS) protection. Firewall Control proxy servers provide methods for SIP signaling and RTP media to traverse firewalls and network address translation (NAT) servers residing at the CSP, ASP, or enterprise. The Core proxy server provides Class 4 switch-like routing capabilities through the NSP's network. These specialized proxy servers, whether deployed as individual or multiple network components, address functional elements of a successful next-generation services network architecture.

Intelligent Opportunities
SIP-centric network architectures also promote the distribution of service intelligence to the terminal devices. PingTel developed a SIP phone that also provides support for Java media and call control programming. Using intelligent terminals such as the PingTel phone, users can create new services by writing their own call handling scripts or by downloading other users' available scripts. Finally, user-installable, fully featured Freeware and Shareware services come to the telephony arena.

Taking this trend to its next logical step is to have the NSPs provide users with complete self-provisioning, accomplished through telephony home pages with links to standard VoiceXML or JTAPI scripts created by the online user community and chargeable shared media services provided by independent CSPs. Imagine creating a script or a service that others can use and getting money off of your phone bill as a result.

The inclusion of SIP-enabled media connectivity in Windows XP creates a huge opportunity for enhanced telephony service self-provisioning. With a SIP-enabled, TAPI-programmable softphone "included" with each PC running Windows XP ("bundled" no longer a Microsoft-approved term), each desktop becomes an intelligent SIP terminal. NSPs concerned with the stickiness of their user-base offerings will adapt to this new user power by offering those media services that are still out of reach of the average user. Access to managed services -- such as large-vocabulary or non-English speech recognition and text-to-speech (TTS), group conferencing, multi-cast low bit-rate streaming, and easy user self-provisioning -- will provide a strong reason for a user to stay with a service provider once his or her personalized environment has been created. In addition to developing SIP-based technologies, Intel has complementary technologies for developing the next generation of IP-based applications and solutions -- most notably the DM3 IPLink development platform.

SIP And IPLink
DM3 IPLink is a standards-based IP-enhanced services platform from Intel Corporation that provides robust, high-performance, next-generation media processing subsystems. The platform is ideal for OEMs, application developers, and IP telephony integrators looking to build next-generation IP telephony solutions for enterprise and service provider markets. Based on modular, flexible architecture, IPLink allows the rapid development of new features and applications, reduces time to market, and increases revenue in addition to ensuring a competitive edge in the next-generation network.

The protocol-agnostic design of IPLink lets developers choose from embedded or host-based signaling protocols. This "split call control" feature allows unprecedented flexibility in implementing next-generation network signaling protocols. The split call control functionality allows an application to directly control the Real Time Protocol (RTP) streaming to and from the IPLink board. Call control is performed in the host application. There are two benefits to host-based call control: First, all the IPLink boards in the system can be addressed using a single IP address for call control. (Separate IP addresses and interfaces are used for the media streaming.) Second, developers can choose from one or more implementations of standard protocols like SIP or MEGACO, or choose a non-standard or proprietary protocol.

Like SIP, IPLink also offers simplicity. Via split call control, this single board PCI or cPCI product provides maximum flexibility, supporting standard IP call control and media gateway protocols such as SIP, H.323, MGCP, and H.248, in addition to a wide variety of vocoder algorithms. Together, SIP and the IPLink platform provide a streamlined and powerful foundation for developing the ground-breaking, next-generation network services that are essential to the success of today's ASPs, CSPs, and NSPs.

Dwight Irving is manager, Application Design Center, Intel Corporation. Mark Manto, product line manager, IP Telephony Products, also contributed to this article.

[ Return To The September 2001 Table Of Contents ]

In a SIP environment, an application server hosts third-party applications such as instant messaging, call control, and tracking user presence on the network. Developers access the application server through APIs and Service Creation Environments (SCEs), and one of the advantages of SIP is its text-based format and its similarity to HTML and other markup languages. Both of these factors contribute to make SIP especially comfortable for Web developers.

Here is a quick list of vendors with products in the SIP application server space.

[ Return To The September 2001 Table Of Contents ]