I Bring in the voice server

Technological shifts have often brought discontinuities to the lives of service providers. The past few years have been no different, filled with hype and uncertainty over asynchronous transfer mode, frame relay and Internet protocol. Is this the year that will bring closure to the heated debate about running voice over any of these technologies?

As networks evolve, start-up carriers are developing at the speed of light, while incumbent carriers are redefining themselves. It is a race against time, and the stakes are high. Who will deploy the best networks? Will they use ATM? Will they use IP? Or will they use IP over ATM? (Telephony, Feb. 1, page 22).

The common-sense answer to this question is: No matter what the transport technology, end users will still expect the same reliability and services they have today.

Voice over ATM or voice over IP is being pursued aggressively; however, few vendors are offering real services. A business case based on cost reductions can be made for implementing either technology using a single network to carry both data and voice. This results in the elimination of overlay networks and simplification of operations.

So where are the services?

Will end users be happy that a service provider is cutting costs if they can only hear one word out of three? Hardly. The technology battle has been won, there is little question of that. Most service providers are building or migrating to ATM networks that will be used for both data and voice services. There is also little question that ATM switches make good sense in a traditional Class 4 or tandem switching environment.

The challenge now is to:

- Interface with existing central offices via the multitude of existing protocols.

- Offer the same toll-quality service for voice.

- Offer the same or improved services such as enhanced free phone or prepaid services.

- Offer the same reliability as the public network.

While the first two points are the main topic of discussion in standards forums, little is being done on the services side.

However, there are ways to handle a smooth integration of the new fabric into the "old skin" of the public network. Advanced Intelligent Network concepts will allow service providers to benefit from the most efficient transport mechanisms as well as a ubiquitous service creation environment.

To make the most out of their networks, service providers will be challenged in several ways. They must interwork with their legacy time division multiplexing (TDM) switches; manage both TDM and ATM networks (or IP networks); and create and offer the same service on both transport fabrics.

In this context, AIN concepts can play a major role in marrying the old with the new by using the same service control points (SCPs) and service creation environment (SCE) for voice services on the TDM fabric as well as the new fabric. By using the SCP to control an ATM network (Figure 1), the ATM switch can play the role of a service switching point (SSP) within an AIN model. However, this puts tremendous pressure on ATM switch providers - ATM switching has little in common with the voice call model and even less in common with the AIN call model.

Introducing the voice server

To comply with the AIN call model and the associated call processing, ATM networks must evolve to become, in some sense, programmable switches that release control of their resources to an external host. To support this new role, a new network element is required: the voice server. The voice server represents the only new element that carriers will require in otherwise established networks.

The voice server can provide control over voice-associated resources, the call model and processing capabilities, and SS7 interfacing with the public network. In this model, an entire ATM network can provide traditional voice services using a single voice control element. In the SS7 network, this is represented as a single point code, making this architecture even more attractive. A simple call flow is described in Figure 2.

This call flow demonstrates how ATM can become the network fabric of choice. Together with the voice server element, ATM can evolve into the new generation of long-distance networks.

Moving a step further, the voice server can provide all the features and capabilities found today in traditional tandem switches:

- Flexible number translation allowing for both domestic and international dialing plans.

- N00/888 translations.

- Routing restrictions by class of service, time of day/day of week and calling party category.

- Authorization.

- Screening.

- Tones and announcements.

Call processing can be modeled to follow the AIN call model with built-in triggers to the long-distance SCP. Carriers can use the same tools for service creation and deployment that they have used for traditional voice services such as travel cards and prepaid services. In this case, the voice server must provide standard AIN interfaces. The most commonly deployed AIN interfaces in networks today are AIN 0.1 and AIN 0.2.

With standard AIN interfaces on the voice server, carriers can deploy a solution that will allow them to generate new seamless services. They will be able to comply with FCC requirements such as local number portability and enjoy the efficiencies and cost savings of a packet-switched network.

Several standards bodies are looking to define the interface between the switching plane and the control plane. For example, the ATM Forum's proxy signaling defined in the UNI 4.0 specification, the Internet Engineering Task Force's media gateway control protocol and the Multi Service Forum's standard were created to address interoperability in increasingly complex networks.

The same concepts apply equally to IP networks. However, in an IP world the nomenclature changes to gateways and gatekeepers.

Migration to next gen networks

The development of an AIN call model to allow call control over any type of switched network (TDM, ATM, IP or wireless) and carrier-grade scalability will make the migration to new technologies and next generation networks much easier.

The beauty of this solution is its simplicity. However, a voice server that provides the required functions to an ATM network must also include scalability in the number of busy-hour call attempts; SS7 links; network management integration; call detail record generation; switchover capabilities (redundancy); and reliability.

All these issues must be addressed sooner rather than later. In addition, a true carrier network must support thousands to millions of busy-hour call attempts. Hardware options are available to increase the needed processing power somewhat. Redundant and fault-tolerant platforms are also available to address the reliability and redundancy issues.

The number of supported links can be easily increased with the choice of a signal transfer point capable of concentrating a very large number of SS7 links and providing SS7 over IP connectivity to the voice server. Higher-density SS7 host-based interfaces are also becoming accepted in the carrier community because they now can support fault-tolerant SS7 architecture and higher connectivity in a very cost-effective solution.

Voice servers are emerging in different forms and shapes: call agents, gatekeepers and call servers.

For voice servers to be able to play a significant role in a carrier network they will need to follow some simple architectural drivers. First, they must have a clear separation of the call control plane and the switching plane. They must also have a clear separation of the call control plane and the services plane.

Voice servers must follow the AIN call model and provide triggering capability to the SCP. They must have a single service creation point independent of the switching fabric and a single point code in the SS7 network. Finally, successful voice servers must be scalable.

As with AIN, the connection between voice servers and ATM or IP fabric will come in many flavors. It may take time for standards to be cast in stone, but one fact is certain: In this age of entrepreneurs, time to market is much more important than waiting for all the standards to be completed.

The concept of using AIN for packet networks has proved to be feasible, while the benefits are clear and attainable. The new voice server offers a solution to the problem and fills an immediate need.